running predictions using NoCodeSeg model about fast-pathology

[smistad]

Sure @andreped , just give me a file which you know causes a crash and I can take a look at it next week

[andreped]

@pr4deepr Just released a patch, which contains a fix for this issue:
https://github.com/AICAN-Research/FAST-Pathology/releases/tag/v0.2.1

Closing this issue for now, as I believe it has been solved.

If you come across any other potential bugs or issues, feel free to report a new issue :]

Thank you for improving the software!

[smistad]

The out of bounds stretch should be fixed in this commit: smistad/FAST@01d1e59

[smistad]

This should fix patches not being processed at left and top edges: smistad/FAST@e8ad46f
Also it should fix an incorrect offset of patches when patch overlap is set in generator.

I am not able to test every case, so beware of bugs. I hope you can test it more thoroughly on your inference cases @andreped

[pr4deepr]

Thanks for the prompt and detailed resposne. I'll try your suggestions tomorrow.
Quick comment that I exported the image as a pyramidal tiff in FIJI (ome-tiff) and in Qupath (tiff). You answered the ome-tiff problem, but am not sure what happened with Qupath exported pyramidal tiff. I'll have a better look tomorrow.
Perhaps not a big deal in pathology, but in microscopy as far as I know, most pyramidal tiff images are saved in ome-tiff formats.

[pr4deepr]

I really wanted to try the WSIs from IBDColEpi dataset but still having trouble downloading for some reason. I've been trying from my university network. I may try downloading from home and see if it works..

Is it possible to upload a single image I can test?

[andreped]

Is it possible to upload a single image I can test?

I uploaded a random selection of WSIs to my personal Google Drive:
https://drive.google.com/drive/folders/1F591dj-d2xAXIMsKqm_qREHoaLEUDPW4?usp=sharing

I just tested the released version on my local machine, and it run without any issues. This is how the prompt should look like (at the end when exporting):

INFO [32300] EXECUTING TIFFImagePyramidExporter because PO is modified.
INFO [32300] Data given to TIFFImagePyramidExporter was an Image, not an ImagePyramid, converting ...
INFO [32300] Processing level 0
INFO [32300] WSI level size: 6720, 4032, 1
INFO [32300] WSI level size: 25.8398 MBs
INFO [32300] Done creating level 0

But I did not use the model that was uploaded the github. I will test that as well. However, it is strange that it crashes when exporting.

You were using Windows 10, right? Not Ubuntu Linux or macOSX? I was testing on Win10. Should work fine there. Tested without making a Project.

[pr4deepr]

HI @andreped
Thanks for sharing the files.
You are right, it works on the large ndpi files you shared, but it is still crashing on the small images I've been trying.
example that worked: 103_HE_inactive.ndpi

Also worked on ID-30_HE_inactive.ndpi

[andreped]

You are right, it works on the large ndpi files you shared, but it is still crashing on the small images I've been trying.

Great! That makes sense. But for the smaller images I am seeing the same on my machine. So there is something wrong with the exporter.

@smistad do you have time to look into it? The TIFFImagePyramidExporter seem to crash on smaller images, likely when it is trying to create the pyramidal TIFF (if the intial segmentation wasn't). You can use the beforementioned WSIs when testing.

[smistad]

Thanks

I believe this fix should resolve this issue: smistad/FAST@e43cc76

@andreped trigger FP v0.2.1 build on github?

[smistad]

Looking at this image, I believe this might be an out-of-bounds issue. If it was a spacing issue, it should gradually be worse throughout the image, but here it is only stretched at the tissue which is very close to the image border (bottom and right).

So if your neural network model uses 512x512 patches, but there is only 400 pixels left at the edge, I'm guessing FAST will resize those 400 pixels to 512, do the segmentation, and then insert the 400 pixels into the result image... What FAST probably should do instead is to pad the 400 pixels to 512 first in the patch generator.

[andreped]

From the most recent commits, the main problems observed in this issue has been solved. What remains is handling the out-of-bounds issue, which should be quick to fix.

But since the original and main issues have been solved, I am closing this issue for now.

Feel free to open a new issue if you find any new bugs or similar :]

[pr4deepr]

Hi
I also tried this on a local Windows PC and go tthe same error:

Selected file: D:/Pradeep/colon.tif
JPEGFixupTagsSubsamplingSec: Warning, Auto-corrected former TIFF subsampling values [2,2] to match subsampling values inside JPEG compressed data [2,1].
count:0
calling insert renderer
finished insert renderer
Model name in wrapper: Epithelium_HE_UNET_512_1P16B32FD6
Final model metadata config sent to pixelClassifier:
m[IE] = (OpenVINO)
m[batch_process] = (2)
m[batch_size] = (1)
m[class_colors] = (0,0,255;0,255,0)
m[class_names] = (Exterior;Epithelium)
m[cpu] = (0)
m[input_img_size_x] = (512)
m[input_img_size_y] = (512)
m[input_node] = (ImageInputLayer)
m[interpolation] = (0)
m[magnification_level] = (10)
m[mask_threshold] = (0.02)
m[model_name] = (Epithelium_HE_512)
m[name] = (Epithelium_HE_512)
m[nb_channels] = (3)
m[nb_classes] = (2)
m[output_node] = (Softmax_Layer_Transpose2)
m[patch_overlap] = (0.05)
m[pipeline] = (import;tissue_segmentation;batchgen;neural_network;stitch;render)
m[problem] = (segmentation)
m[resolution] = (high)
m[scale_factor] = (1.0f/1.0f)
m[task] = (Epithelium_HE_512)
m[tissue_threshold] = (70)
Final model metadata config WITHIN to pixelClassifier:
m[IE] = (OpenVINO)
m[batch_process] = (2)
m[batch_size] = (1)
m[class_colors] = (0,0,255;0,255,0)
m[class_names] = (Exterior;Epithelium)
m[cpu] = (0)
m[input_img_size_x] = (512)
m[input_img_size_y] = (512)
m[input_node] = (ImageInputLayer)
m[interpolation] = (0)
m[magnification_level] = (10)
m[mask_threshold] = (0.02)
m[model_name] = (Epithelium_HE_512)
m[name] = (Epithelium_HE_512)
m[nb_channels] = (3)
m[nb_classes] = (2)
m[output_node] = (Softmax_Layer_Transpose2)
m[patch_overlap] = (0.05)
m[pipeline] = (import;tissue_segmentation;batchgen;neural_network;stitch;render)
m[problem] = (segmentation)
m[resolution] = (high)
m[scale_factor] = (1.0f/1.0f)
m[task] = (Epithelium_HE_512)
m[tissue_threshold] = (70)
Current model: Epithelium_HE_UNET_512_1P16B32FD6
current WSI: D:/Pradeep/colon.tif
ERROR [19076] Std exception caught in Qt event handler invalid stof argument
ERROR [19076] Std exception caught in Qt event handler invalid stof argument
ERROR [19076] Terminated with unhandled exception: invalid stof argument

Intel Xeon CPU E5-2640
NVIDIA Quardro P4000

[andreped]

You came to the right place, @pr4deepr.

Sad to hear that you are having issues.

Regarding OME-TIFF, OpenSlide does not currently support this format, which is why you are unable to read this image in FastPathology. However, how were you able to choose this image? Did you use the "Import WSI" button or did you drag-and-drop? In theory, you should only be able to choose between supported formats:
https://github.com/AICAN-Research/FAST-Pathology/blob/master/source/MainWindow.cpp#L1470

If you are indeed able to, somehow, it is true that the program might crash. I will look into making selection of files more stable. New release right around the corner! Hopefully before christmas.

Regarding deployment with the ONNX model on the TIFF image you were trying to use, which TIFF format is this? Generic TIFF? I just noticed that there was an interesting prompt:
"JPEGFixupTagsSubsamplingSec: Warning, Auto-corrected former TIFF subsampling values [2,2] to match subsampling values inside JPEG compressed data [2,1].

I have seen this before, when I was converting WSIs stored in the CellSens VSI format -> BigTIFF and used JPEG compression using the command line tool vips. Perhaps that is the reason. However, I believe you were able to select and visualize the WSI, as you got all the way to running inference, right?

One thing you could try is to run FastPathology from the shell. Are you using Windows? If yes, assuming that you installed FastPathology at the default location, you can try running this command using PowerShell:
& C:\Program Files\FastPathology\bin\fastpathology.exe --verbose

Essentially running the software in verbose mode, which may make debugging a little easier (should be more info). Try redoing what you did and see what it prompts. Let me know how it goes. I mostly tested the software on Windows at the end. Might be that there is something that is not compatible with for instance Linux, but that should be easy to fix. macOSX support is quite experimental.

[pr4deepr]

I'm using Windows 10 at the moment.
I've run FastPathology in verbose mode and this is the output

INFO [1880] Loaded configuration file: C:/Program Files/FastPathology/bin/fast_configuration.txt
INFO [1880] Test data path: C:/ProgramData/FAST/data/
INFO [1880] Kernel source path: C:/Program Files/FastPathology/bin//..//kernels/
INFO [1880] Kernel binary path: C:/ProgramData/FAST/kernel_binaries/
INFO [1880] Documentation path: C:/Program Files/FastPathology/bin//..//doc/
INFO [1880] Pipeline path: C:/ProgramData/FAST/pipelines/
INFO [1880] Qt plugins path: C:/Program Files/FastPathology/bin//..//plugins/
INFO [1880] Library path: C:/Program Files/FastPathology/bin//..//bin/
INFO [1880] Creating new QApp
INFO [1880] Creating new GL context for computation thread
INFO [1880] Medium large screen detected with width: 2560
Temporary path (UPDATED!): C:/Users/rajasekhar.p/AppData/Local/Temp/fastpathology-wzzcGf
Does folder exist:1
Current temporary project folder location:C:/Users/rajasekhar.p/AppData/Local/Temp/fastpathology-wzzcGf/project_17409488
INFO [1880] QApp already exists..
INFO [1880] Found 1 OpenCL platforms.
INFO [1880] 1 platforms selected for inspection.
INFO [1880] Platform 0: NVIDIA Corporation
INFO [1880] This platform has 1 available devices in total
INFO [1880] Looking for GPU devices only.
INFO [1880] 1 selected.
INFO [1880] Inspecting device 0 with the name Quadro P4000
INFO [1880] There are 1 devices that can be associated with the GL context
INFO [1880] 000002887D462E10 - 000002887D462E10
INFO [1880] Device has OpenGL interop capability
INFO [1880] The device was accepted.
INFO [1880] Quadro P4000 has 14 compute units
INFO [1880] The device Quadro P4000 got a score of 1000014
INFO [1880] The platform NVIDIA CUDA was selected as the best platform.
INFO [1880] A total of 1 devices were selected for the context from this platform:
INFO [1880] The best device was: Quadro P4000
INFO [1880] The following device was selected as main device: Quadro P4000
INFO [1880] Writing directly to 3D textures/images is NOT supported on main device
INFO [1880] Resizing window to 1024 1024
INFO [1880] Window not minimized; turning ON synchronized rendering
INFO [1880] Trying to start computation thread
INFO [1880] Computation thread started
Selected file: D:/Pradeep/colon.tif
INFO [1880] EXECUTING WholeSlideImageImporter because PO is modified.
JPEGFixupTagsSubsamplingSec: Warning, Auto-corrected former TIFF subsampling values [2,2] to match subsampling values inside JPEG compressed data [2,1].
INFO [1880] Metadata: openslide.comment =
10240x12288 (256x256) JPEG/RGB Q=80|Date = 26/11/2021|Time = 15:56:12|MPP = 0.24984|3dh_PixelSizeX = 0.249711|3dh_PixelSizeY = 0.249968|3dh_Filter = Default|3dh_Profile = Routine x20|3dh_ScannerHardwareID = SC150-210905
INFO [1880] Metadata: openslide.level-count = 1
INFO [1880] Metadata: openslide.level[0].downsample = 1
INFO [1880] Metadata: openslide.level[0].height = 12288
INFO [1880] Metadata: openslide.level[0].tile-height = 256
INFO [1880] Metadata: openslide.level[0].tile-width = 256
INFO [1880] Metadata: openslide.level[0].width = 10240
INFO [1880] Metadata: openslide.vendor = generic-tiff
INFO [1880] Metadata: tiff.ImageDescription =
10240x12288 (256x256) JPEG/RGB Q=80|Date = 26/11/2021|Time = 15:56:12|MPP = 0.24984|3dh_PixelSizeX = 0.249711|3dh_PixelSizeY = 0.249968|3dh_Filter = Default|3dh_Profile = Routine x20|3dh_ScannerHardwareID = SC150-210905
INFO [1880] Metadata: tiff.ResolutionUnit = centimeter
INFO [1880] Metadata: tiff.XResolution = 40046.315104895104
INFO [1880] Metadata: tiff.YResolution = 40005.284044336811
INFO [1880] WSI has 1 levels
INFO [1880] Checking level 0
INFO [1880] WSI level 0 has size 10240x12288 and 480 MB adding..
INFO [1880] Setting tile width and height to 256 256
count:0
calling insert renderer
INFO [1880] EXECUTING EmptyProcessObject because PO is modified.
INFO [1880] EXECUTING ImagePyramidRenderer because PO is modified.
INFO [1880] Recalculating the camera of the view
INFO [1880] Finished initializing OpenGL
finished insert renderer
INFO [1880] Recalculating the camera of the view
INFO [1880] Finished initializing OpenGL
INFO [1880] EXECUTING EmptyProcessObject because PO is modified.
INFO [1880] EXECUTING ImageChannelConverter because PO is modified.
Model name in wrapper: Epithelium_HE_UNET_512_1P16B32FD6
Final model metadata config sent to pixelClassifier:
m[IE] = (OpenVINO)
m[batch_process] = (2)
m[batch_size] = (1)
m[class_colors] = (0,0,255;0,255,0)
m[class_names] = (Exterior;Epithelium)
m[cpu] = (0)
m[input_img_size_x] = (512)
m[input_img_size_y] = (512)
m[input_node] = (ImageInputLayer)
m[interpolation] = (0)
m[magnification_level] = (10)
m[mask_threshold] = (0.02)
m[model_name] = (Epithelium_HE_512)
m[name] = (Epithelium_HE_512)
m[nb_channels] = (3)
m[nb_classes] = (2)
m[output_node] = (Softmax_Layer_Transpose2)
m[patch_overlap] = (0.05)
m[pipeline] = (import;tissue_segmentation;batchgen;neural_network;stitch;render)
m[problem] = (segmentation)
m[resolution] = (high)
m[scale_factor] = (1.0f/1.0f)
m[task] = (Epithelium_HE_512)
m[tissue_threshold] = (70)
Final model metadata config WITHIN to pixelClassifier:
m[IE] = (OpenVINO)
m[batch_process] = (2)
m[batch_size] = (1)
m[class_colors] = (0,0,255;0,255,0)
m[class_names] = (Exterior;Epithelium)
m[cpu] = (0)
m[input_img_size_x] = (512)
m[input_img_size_y] = (512)
m[input_node] = (ImageInputLayer)
m[interpolation] = (0)
m[magnification_level] = (10)
m[mask_threshold] = (0.02)
m[model_name] = (Epithelium_HE_512)
m[name] = (Epithelium_HE_512)
m[nb_channels] = (3)
m[nb_classes] = (2)
m[output_node] = (Softmax_Layer_Transpose2)
m[patch_overlap] = (0.05)
m[pipeline] = (import;tissue_segmentation;batchgen;neural_network;stitch;render)
m[problem] = (segmentation)
m[resolution] = (high)
m[scale_factor] = (1.0f/1.0f)
m[task] = (Epithelium_HE_512)
m[tissue_threshold] = (70)
Current model: Epithelium_HE_UNET_512_1P16B32FD6
current WSI: D:/Pradeep/colon.tif
ERROR [1880] Std exception caught in Qt event handler invalid stof argument
ERROR [1880] Std exception caught in Qt event handler invalid stof argument
ERROR [1880] Terminated with unhandled exception: invalid stof argument

In answer to your questions.

. However, how were you able to choose this image? Did you use the "Import WSI" button or did you drag-and-drop?

I used the Import WSI button and thats when it crashed.

I have seen this before, when I was converting WSIs stored in the CellSens VSI format -> BigTIFF and used JPEG compression using the command line tool vips. Perhaps that is the reason. However, I believe you were able to select and visualize the WSI, as you got all the way to running inference, right?
Yea, I can open the image as displayed below:

When I run the inference, it crashes..
The original file is mrxs format which I convert to tiff. In this case I've been using the software, Caseviewer.

The other option is running inference on QuPath. Is it possible to get the model file for running on QuPath? I assume the ones on NoCodeSeg are for FastPathology?

[pr4deepr]

Sorry, I meant is it possible to get the model to run in MIB... not qupath..

[andreped]

Inference in QuPath:
Inference support in QuPath is still a work-in-progress. It does not currently supports deployment of pretrained CNNs, at least not to the general public. I know this is something Peter Bankhead is working on, but there is still some work to be done. I look forward to deployment support in QuPath myself :]

Inference in MIB:
An alternative is to use MIB, as mentioned in the code-free manuscript. However, FastPathology is the best solution for deployment, both in terms of speed, generalizability, and supports visualization of the WSI with segmentation overlay, whereas MIB only works with patches (if you also wish to render predictions with the image). But for pure inference, that is apply trained model on data and store results on disk, to be visualized and processed further elsewhere is possible in MIB. See the tutorial video of how this was done. However, I would recommend using FastPathology for deployment in general.

Inference bug in FastPathology:
Oh, I think I see whats wrong. Your WSI is only stored as a TIFF, but not as a tiled, pyramidal TIFF. From the verbose prompt you can see that openslide.level-count = 1. You can also see openslide.level[0].height = 12288, but no mention of openslide Hence, only image plane 0 exists. From reading the tiff.XResolution = 40046.315104895104, I assume that the WSI scanned at x40 magnification? The ONNX model is to be applied on the x10 magnification image plane (that is downsample = 4, assuming a 40x WSI). FastPathology attempts to look for this image plane (plane=2), but fails, as there only exists image plane 0.

I should add a check for this in the near future. Thank you for spotting this bug! Likely I will prompt the user that the optimal image plane was not found and an alternative was chosen instead.

Quick-fix:
If you just want to see that the ONNX model runs on your image, setting the "magnification_level" attribute in the model .txt-file to "40" instead of "10" should likely solve your issue. However, segmentation performance will likely be worse, as it is not trained at this resolution.

To change this value, you could do this by clicking the button that says "Clinical mode", button bottom-left, which should then toggle to "Research mode". When you then click inference for the ONNX model, a small dialog should be displayed, where you can change the hyperparameters as you wish (NOTE: Will crash if you add something silly - to be made more stable in the future). Note that this only temporarily changes the value, just for this run, and does not update the actual model configuration file (.txt).

Alternatively, you could modify the configuration .txt-file directly. Since you are on Windows, after the models have been added, they should be located in you user space. For instance: "C:\Users\andrp\fastpathology\data\Models\some-model-file.txt". Then you can just modify the specific .txt-file you want to update. This updates the .txt-file directly. FastPathology also includes a simple script/text/pipeline editor, which you could use to do this, but feel free to use whichever editor you prefer.

Converting WSIs to tiled, pyramidal TIFFs:
As FastPathology assumes that the input image is stored in this tiled, pyramidal format, I would recommend converting to this. It does not take that much extra space on disk, and other softwares like QuPath also benefit from working with pyramidal images, compared to naive, ordinary and large TIFFs.

To convert your TIFF image to tiled, pyramidal TIFF, you can see how I did this in my vsi2tif converter. If you do not wish to use python, you can use the vips command line tool directly.

To convert from TIFF to tiled, pyramidal TIFF, you could do this:

vips tiffsave path-to-regular-input-image.tif path-to-pyramidal-output-image.tif --tile --pyramid --compression=JPEG --Q=85

You might have to add the --bigtiff flag, if your image is saved in this format. If you get any issues with JPEG, which I guess you might get, you could attempt to decompress the image beforehand, before doing the conversion. For that I recommend using ImageMagick. See here for an example of how to do this.

Hope this helps! Be sure to let me know if changing the configuration file helped, or if you get any other issues. Happy to help :]

[andreped]

Hmm, might be that FIJI does not create image planes if they are very small. What you often do is that you create image planes that are downsampled versions of the original image, for instance by 2 or 4. Might be that they just neglected those smaller ones. Not sure.

But yeah, I gave you some ideas of how you could do this conversion. It is quite straight forward.

BTW: What was the original image format? OpenSlide supports a wide range of different formats. Might be that you could use the original format with FastPathology directly.

BTW2: The annotated data set which we used in the code-free study will be made openly available today. I can send you the link here, if you wish to download the data. At least you could test those WSIs, that you are able to run inference on those (as a sanity check).

Good luck!

[andreped]

The data set was published on dataverseNO just now. It is completely open. You can download some of the WSIs we used for testing FastPathology, if you see any odd behaviour or if you are still having issues tomorrow (for sanity checking). The data set can be accessed from here.

[pr4deepr]

@andreped Thanks so much!
This WSI dataset is very valuable. Appreciate you making this available for everyone...

So, I performed the conversion using vips command line:
.\vips.exe tiffsave D:\Pradeep\confocal.tif D:\Pradeep\Andre\pyramidal_conv\pyramidal.tif --tile --pyramid --compression=jpeg --Q=85

with and without the bigtiff flag and now I'm getting another error when I run inference:

INFO [7736] Loaded configuration file: C:/Program Files/FastPathology/bin/fast_configuration.txt
INFO [7736] Test data path: C:/ProgramData/FAST/data/
INFO [7736] Kernel source path: C:/Program Files/FastPathology/bin//..//kernels/
INFO [7736] Kernel binary path: C:/ProgramData/FAST/kernel_binaries/
INFO [7736] Documentation path: C:/Program Files/FastPathology/bin//..//doc/
INFO [7736] Pipeline path: C:/ProgramData/FAST/pipelines/
INFO [7736] Qt plugins path: C:/Program Files/FastPathology/bin//..//plugins/
INFO [7736] Library path: C:/Program Files/FastPathology/bin//..//bin/
INFO [7736] Creating new QApp
INFO [7736] Creating new GL context for computation thread
INFO [7736] Medium large screen detected with width: 2560
Temporary path (UPDATED!): C:/Users/rajasekhar.p/AppData/Local/Temp/fastpathology-uGVLnu
Does folder exist:1
Current temporary project folder location:C:/Users/rajasekhar.p/AppData/Local/Temp/fastpathology-uGVLnu/project_17409488
INFO [7736] QApp already exists..
INFO [7736] Found 1 OpenCL platforms.
INFO [7736] 1 platforms selected for inspection.
INFO [7736] Platform 0: NVIDIA Corporation
INFO [7736] This platform has 1 available devices in total
INFO [7736] Looking for GPU devices only.
INFO [7736] 1 selected.
INFO [7736] Inspecting device 0 with the name Quadro P4000
INFO [7736] There are 1 devices that can be associated with the GL context
INFO [7736] 000001E1FB33DAD0 - 000001E1FB33DAD0
INFO [7736] Device has OpenGL interop capability
INFO [7736] The device was accepted.
INFO [7736] Quadro P4000 has 14 compute units
INFO [7736] The device Quadro P4000 got a score of 1000014
INFO [7736] The platform NVIDIA CUDA was selected as the best platform.
INFO [7736] A total of 1 devices were selected for the context from this platform:
INFO [7736] The best device was: Quadro P4000
INFO [7736] The following device was selected as main device: Quadro P4000
INFO [7736] Writing directly to 3D textures/images is NOT supported on main device
INFO [7736] Resizing window to 1024 1024
INFO [7736] Window not minimized; turning ON synchronized rendering
INFO [7736] Trying to start computation thread
INFO [7736] Computation thread started
INFO [7736] Drag event received in window widget
INFO [7736] Drop event received in window widget
INFO [7736] Dropped file:D:/Pradeep/Andre/pyramidal_conv/pyramidal.tif
Selected file: D:/Pradeep/Andre/pyramidal_conv/pyramidal.tif
INFO [7736] EXECUTING WholeSlideImageImporter because PO is modified.
INFO [7736] Metadata: openslide.comment =
10240x12288 (256x256) JPEG/RGB Q=80|Date = 26/11/2021|Time = 15:56:12|MPP = 0.24984|3dh_PixelSizeX = 0.249711|3dh_PixelSizeY = 0.249968|3dh_Filter = Default|3dh_Profile = Routine x20|3dh_ScannerHardwareID = SC150-210905
INFO [7736] Metadata: openslide.level-count = 8
INFO [7736] Metadata: openslide.level[0].downsample = 1
INFO [7736] Metadata: openslide.level[0].height = 12288
INFO [7736] Metadata: openslide.level[0].tile-height = 128
INFO [7736] Metadata: openslide.level[0].tile-width = 128
INFO [7736] Metadata: openslide.level[0].width = 10240
INFO [7736] Metadata: openslide.level[1].downsample = 2
INFO [7736] Metadata: openslide.level[1].height = 6144
INFO [7736] Metadata: openslide.level[1].tile-height = 128
INFO [7736] Metadata: openslide.level[1].tile-width = 128
INFO [7736] Metadata: openslide.level[1].width = 5120
INFO [7736] Metadata: openslide.level[2].downsample = 4
INFO [7736] Metadata: openslide.level[2].height = 3072
INFO [7736] Metadata: openslide.level[2].tile-height = 128
INFO [7736] Metadata: openslide.level[2].tile-width = 128
INFO [7736] Metadata: openslide.level[2].width = 2560
INFO [7736] Metadata: openslide.level[3].downsample = 8
INFO [7736] Metadata: openslide.level[3].height = 1536
INFO [7736] Metadata: openslide.level[3].tile-height = 128
INFO [7736] Metadata: openslide.level[3].tile-width = 128
INFO [7736] Metadata: openslide.level[3].width = 1280
INFO [7736] Metadata: openslide.level[4].downsample = 16
INFO [7736] Metadata: openslide.level[4].height = 768
INFO [7736] Metadata: openslide.level[4].tile-height = 128
INFO [7736] Metadata: openslide.level[4].tile-width = 128
INFO [7736] Metadata: openslide.level[4].width = 640
INFO [7736] Metadata: openslide.level[5].downsample = 32
INFO [7736] Metadata: openslide.level[5].height = 384
INFO [7736] Metadata: openslide.level[5].tile-height = 128
INFO [7736] Metadata: openslide.level[5].tile-width = 128
INFO [7736] Metadata: openslide.level[5].width = 320
INFO [7736] Metadata: openslide.level[6].downsample = 64
INFO [7736] Metadata: openslide.level[6].height = 192
INFO [7736] Metadata: openslide.level[6].tile-height = 128
INFO [7736] Metadata: openslide.level[6].tile-width = 128
INFO [7736] Metadata: openslide.level[6].width = 160
INFO [7736] Metadata: openslide.level[7].downsample = 128
INFO [7736] Metadata: openslide.level[7].height = 96
INFO [7736] Metadata: openslide.level[7].tile-height = 128
INFO [7736] Metadata: openslide.level[7].tile-width = 128
INFO [7736] Metadata: openslide.level[7].width = 80
INFO [7736] Metadata: openslide.quickhash-1 = f65e4e1293996bbe2c75dbb9b5012169822d702f8fdd9210540bd265237bb942
INFO [7736] Metadata: openslide.vendor = generic-tiff
INFO [7736] Metadata: tiff.ImageDescription =
10240x12288 (256x256) JPEG/RGB Q=80|Date = 26/11/2021|Time = 15:56:12|MPP = 0.24984|3dh_PixelSizeX = 0.249711|3dh_PixelSizeY = 0.249968|3dh_Filter = Default|3dh_Profile = Routine x20|3dh_ScannerHardwareID = SC150-210905
INFO [7736] Metadata: tiff.ResolutionUnit = centimeter
INFO [7736] Metadata: tiff.XResolution = 40046.31640625
INFO [7736] Metadata: tiff.YResolution = 40005.28515625
INFO [7736] WSI has 8 levels
INFO [7736] Checking level 0
INFO [7736] WSI level 0 has size 10240x12288 and 480 MB adding..
INFO [7736] Setting tile width and height to 128 128
INFO [7736] Checking level 1
INFO [7736] WSI level 1 has size 5120x6144 and 120 MB adding..
INFO [7736] Setting tile width and height to 128 128
INFO [7736] Checking level 2
INFO [7736] WSI level 2 has size 2560x3072 and 30 MB adding..
INFO [7736] Setting tile width and height to 128 128
INFO [7736] Checking level 3
INFO [7736] WSI level 3 has size 1280x1536 and 7 MB adding..
INFO [7736] Setting tile width and height to 128 128
INFO [7736] Checking level 4
INFO [7736] WSI level was less than 4 MB, skipping..
count:0
calling insert renderer
INFO [7736] EXECUTING EmptyProcessObject because PO is modified.
INFO [7736] EXECUTING ImagePyramidRenderer because PO is modified.
INFO [7736] Recalculating the camera of the view
INFO [7736] Finished initializing OpenGL
finished insert renderer
INFO [7736] Recalculating the camera of the view
INFO [7736] Finished initializing OpenGL
INFO [7736] EXECUTING EmptyProcessObject because PO is modified.
INFO [7736] EXECUTING ImageChannelConverter because PO is modified.
Model name in wrapper: Epithelium_HE_UNET_512_1P16B32FD6
Final model metadata config sent to pixelClassifier:
m[IE] = (TensorRt)
m[batch_process] = (2)
m[batch_size] = (1)
m[class_colors] = (0,0,255;0,255,0)
m[class_names] = (Exterior;Epithelium)
m[cpu] = (0)
m[input_img_size_x] = (512)
m[input_img_size_y] = (512)
m[input_node] = (ImageInputLayer)
m[interpolation] = (0)
m[magnification_level] = (10)
m[mask_threshold] = (0.02)
m[model_name] = (Epithelium_HE_512)
m[name] = (Epithelium_HE_512)
m[nb_channels] = (3)
m[nb_classes] = (2)
m[output_node] = (Softmax_Layer_Transpose2)
m[patch_overlap] = (0.05)
m[pipeline] = (import;tissue_segmentation;batchgen;neural_network;stitch;render)
m[problem] = (segmentation)
m[resolution] = (high)
m[scale_factor] = (1.0f/1.0f)
m[task] = (Epithelium_HE_512)
m[tissue_threshold] = (70)
Final model metadata config WITHIN to pixelClassifier:
m[IE] = (TensorRt)
m[batch_process] = (2)
m[batch_size] = (1)
m[class_colors] = (0,0,255;0,255,0)
m[class_names] = (Exterior;Epithelium)
m[cpu] = (0)
m[input_img_size_x] = (512)
m[input_img_size_y] = (512)
m[input_node] = (ImageInputLayer)
m[interpolation] = (0)
m[magnification_level] = (10)
m[mask_threshold] = (0.02)
m[model_name] = (Epithelium_HE_512)
m[name] = (Epithelium_HE_512)
m[nb_channels] = (3)
m[nb_classes] = (2)
m[output_node] = (Softmax_Layer_Transpose2)
m[patch_overlap] = (0.05)
m[pipeline] = (import;tissue_segmentation;batchgen;neural_network;stitch;render)
m[problem] = (segmentation)
m[resolution] = (high)
m[scale_factor] = (1.0f/1.0f)
m[task] = (Epithelium_HE_512)
m[tissue_threshold] = (70)
Current model: Epithelium_HE_UNET_512_1P16B32FD6
current WSI: D:/Pradeep/Andre/pyramidal_conv/pyramidal.tif
Curr patch level: 2
current WSI: D:/Pradeep/Andre/pyramidal_conv/pyramidal.tif
INFO [7736] EXECUTING WholeSlideImageImporter because PO is modified.
INFO [7736] Metadata: openslide.comment =
10240x12288 (256x256) JPEG/RGB Q=80|Date = 26/11/2021|Time = 15:56:12|MPP = 0.24984|3dh_PixelSizeX = 0.249711|3dh_PixelSizeY = 0.249968|3dh_Filter = Default|3dh_Profile = Routine x20|3dh_ScannerHardwareID = SC150-210905
INFO [7736] Metadata: openslide.level-count = 8
INFO [7736] Metadata: openslide.level[0].downsample = 1
INFO [7736] Metadata: openslide.level[0].height = 12288
INFO [7736] Metadata: openslide.level[0].tile-height = 128
INFO [7736] Metadata: openslide.level[0].tile-width = 128
INFO [7736] Metadata: openslide.level[0].width = 10240
INFO [7736] Metadata: openslide.level[1].downsample = 2
INFO [7736] Metadata: openslide.level[1].height = 6144
INFO [7736] Metadata: openslide.level[1].tile-height = 128
INFO [7736] Metadata: openslide.level[1].tile-width = 128
INFO [7736] Metadata: openslide.level[1].width = 5120
INFO [7736] Metadata: openslide.level[2].downsample = 4
INFO [7736] Metadata: openslide.level[2].height = 3072
INFO [7736] Metadata: openslide.level[2].tile-height = 128
INFO [7736] Metadata: openslide.level[2].tile-width = 128
INFO [7736] Metadata: openslide.level[2].width = 2560
INFO [7736] Metadata: openslide.level[3].downsample = 8
INFO [7736] Metadata: openslide.level[3].height = 1536
INFO [7736] Metadata: openslide.level[3].tile-height = 128
INFO [7736] Metadata: openslide.level[3].tile-width = 128
INFO [7736] Metadata: openslide.level[3].width = 1280
INFO [7736] Metadata: openslide.level[4].downsample = 16
INFO [7736] Metadata: openslide.level[4].height = 768
INFO [7736] Metadata: openslide.level[4].tile-height = 128
INFO [7736] Metadata: openslide.level[4].tile-width = 128
INFO [7736] Metadata: openslide.level[4].width = 640
INFO [7736] Metadata: openslide.level[5].downsample = 32
INFO [7736] Metadata: openslide.level[5].height = 384
INFO [7736] Metadata: openslide.level[5].tile-height = 128
INFO [7736] Metadata: openslide.level[5].tile-width = 128
INFO [7736] Metadata: openslide.level[5].width = 320
INFO [7736] Metadata: openslide.level[6].downsample = 64
INFO [7736] Metadata: openslide.level[6].height = 192
INFO [7736] Metadata: openslide.level[6].tile-height = 128
INFO [7736] Metadata: openslide.level[6].tile-width = 128
INFO [7736] Metadata: openslide.level[6].width = 160
INFO [7736] Metadata: openslide.level[7].downsample = 128
INFO [7736] Metadata: openslide.level[7].height = 96
INFO [7736] Metadata: openslide.level[7].tile-height = 128
INFO [7736] Metadata: openslide.level[7].tile-width = 128
INFO [7736] Metadata: openslide.level[7].width = 80
INFO [7736] Metadata: openslide.quickhash-1 = f65e4e1293996bbe2c75dbb9b5012169822d702f8fdd9210540bd265237bb942
INFO [7736] Metadata: openslide.vendor = generic-tiff
INFO [7736] Metadata: tiff.ImageDescription =
10240x12288 (256x256) JPEG/RGB Q=80|Date = 26/11/2021|Time = 15:56:12|MPP = 0.24984|3dh_PixelSizeX = 0.249711|3dh_PixelSizeY = 0.249968|3dh_Filter = Default|3dh_Profile = Routine x20|3dh_ScannerHardwareID = SC150-210905
INFO [7736] Metadata: tiff.ResolutionUnit = centimeter
INFO [7736] Metadata: tiff.XResolution = 40046.31640625
INFO [7736] Metadata: tiff.YResolution = 40005.28515625
INFO [7736] WSI has 8 levels
INFO [7736] Checking level 0
INFO [7736] WSI level 0 has size 10240x12288 and 480 MB adding..
INFO [7736] Setting tile width and height to 128 128
INFO [7736] Checking level 1
INFO [7736] WSI level 1 has size 5120x6144 and 120 MB adding..
INFO [7736] Setting tile width and height to 128 128
INFO [7736] Checking level 2
INFO [7736] WSI level 2 has size 2560x3072 and 30 MB adding..
INFO [7736] Setting tile width and height to 128 128
INFO [7736] Checking level 3
INFO [7736] WSI level 3 has size 1280x1536 and 7 MB adding..
INFO [7736] Setting tile width and height to 128 128
INFO [7736] Checking level 4
INFO [7736] WSI level was less than 4 MB, skipping..
Current OS is: windows
Current kernel is: winnt
accepted models: .onnx
accepted models: .txt
INFO [7736] Loading inference engines in folder C:/Program Files/FastPathology/bin//..//bin/
INFO [7736] Loading inference engine OpenVINO from shared library InferenceEngineOpenVINO.dll
INFO [7736] Loading inference engine TensorFlow from shared library InferenceEngineTensorFlow.dll
2021-12-10 09:38:08.699748: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library cudart64_110.dll
INFO [7736] Loading inference engine TensorRT from shared library InferenceEngineTensorRT.dll
WARNING [7736] Failed to load plugin because The specified module could not be found.

INFO [7736] Inference engine OpenVINO selected
INFO [7736] Inference engine OpenVINO selected
Current available IEs:
OpenVINO,
TensorFlow,
TensorRT,
Which model formats are available and that there exists an IE for:
.onnx,
.txt,
TensorRT (using ONNX) selected
ERROR [7736] FAST exception caught in Qt event handler Inference engine with name TensorRT is not available
ERROR [7736] FAST exception caught in Qt event handler Inference engine with name TensorRT is not available
ERROR [7736] Terminated with unhandled exception: Inference engine with name TensorRT is not available

Copy of the model config file:

model_name:Epithelium_HE_512
name:Epithelium_HE_512
task:Epithelium_HE_512
problem:segmentation
resolution:high
magnification_level:10
mask_threshold:0.02
tissue_threshold:70
patch_overlap:0.05
batch_size:1
input_img_size_x:512
input_img_size_y:512
nb_channels:3
nb_classes:2
input_node:ImageInputLayer
output_node:Softmax_Layer_Transpose2
class_colors:0,0,255;0,255,0
class_names:Exterior;Epithelium
interpolation:0
pipeline:import;tissue_segmentation;batchgen;neural_network;stitch;render
batch_process:2
scale_factor:1.0f/1.0f
cpu:0
IE:OpenVINO

[pr4deepr]

Also, I'm having trouble downloading the data from DataverseNO. It keeps downloading half way and then stops/fails..

[andreped]

Also, I'm having trouble downloading the data from DataverseNO. It keeps downloading half way and then stops/fails..

Did you try downloading everything at once or one file at a time?

At least it worked for me just now downloading the TIFF-annotations.zip and WSI_part_02.zip, separately.

[andreped]

Regarding inference:
I see that you had written 'TensorRt' for the 'IE' attribute in the model config file. First of it is written 'TensorRT'. Can you try setting 'IE' to 'OpenVINO' and running again? Then show the prompt.

Also, do you have CUDA installed? You shouldn't need to, but it might seem like FP thinks you have TensorRT installed if CUDA is installed.

However, later on your code crashed because it tried to use TensorRT which you had not installed (has to be installed separately - same goes for CUDA and cudNN).

[pr4deepr]

Hi @andreped
I didn't realise I had to install TensorRT. I've done that and changed the model config file so its "TensorRT"
I also have CUDA and cudnn installed.

I can the inference till where I can see the green overlay on top of the image, after which fastpathology software just crashes.

INFO [23968] Loaded configuration file: C:/Program Files/FastPathology/bin/fast_configuration.txt
INFO [23968] Test data path: C:/ProgramData/FAST/data/
INFO [23968] Kernel source path: C:/Program Files/FastPathology/bin//..//kernels/
INFO [23968] Kernel binary path: C:/ProgramData/FAST/kernel_binaries/
INFO [23968] Documentation path: C:/Program Files/FastPathology/bin//..//doc/
INFO [23968] Pipeline path: C:/ProgramData/FAST/pipelines/
INFO [23968] Qt plugins path: C:/Program Files/FastPathology/bin//..//plugins/
INFO [23968] Library path: C:/Program Files/FastPathology/bin//..//bin/
INFO [23968] Creating new QApp
INFO [23968] Creating new GL context for computation thread
INFO [23968] Medium large screen detected with width: 2560
Temporary path (UPDATED!): C:/Users/rajasekhar.p/AppData/Local/Temp/fastpathology-MlbLYw
Does folder exist:1
Current temporary project folder location:C:/Users/rajasekhar.p/AppData/Local/Temp/fastpathology-MlbLYw/project_17409488
INFO [23968] QApp already exists..
INFO [23968] Found 1 OpenCL platforms.
INFO [23968] 1 platforms selected for inspection.
INFO [23968] Platform 0: NVIDIA Corporation
INFO [23968] This platform has 1 available devices in total
INFO [23968] Looking for GPU devices only.
INFO [23968] 1 selected.
INFO [23968] Inspecting device 0 with the name Quadro P4000
INFO [23968] There are 1 devices that can be associated with the GL context
INFO [23968] 000001AF17BA1FF0 - 000001AF17BA1FF0
INFO [23968] Device has OpenGL interop capability
INFO [23968] The device was accepted.
INFO [23968] Quadro P4000 has 14 compute units
INFO [23968] The device Quadro P4000 got a score of 1000014
INFO [23968] The platform NVIDIA CUDA was selected as the best platform.
INFO [23968] A total of 1 devices were selected for the context from this platform:
INFO [23968] The best device was: Quadro P4000
INFO [23968] The following device was selected as main device: Quadro P4000
INFO [23968] Writing directly to 3D textures/images is NOT supported on main device
INFO [23968] Resizing window to 1024 1024
INFO [23968] Window not minimized; turning ON synchronized rendering
INFO [23968] Trying to start computation thread
INFO [23968] Computation thread started
INFO [23968] Recalculating the camera of the view
INFO [23968] Drag event received in window widget
INFO [23968] Drop event received in window widget
INFO [23968] Dropped file:D:/Pradeep/Andre/pyramidal_conv/pyramidal.tif
Selected file: D:/Pradeep/Andre/pyramidal_conv/pyramidal.tif
INFO [23968] EXECUTING WholeSlideImageImporter because PO is modified.
INFO [23968] Metadata: openslide.comment =
10240x12288 (256x256) JPEG/RGB Q=80|Date = 26/11/2021|Time = 15:56:12|MPP = 0.24984|3dh_PixelSizeX = 0.249711|3dh_PixelSizeY = 0.249968|3dh_Filter = Default|3dh_Profile = Routine x20|3dh_ScannerHardwareID = SC150-210905
INFO [23968] Metadata: openslide.level-count = 8
INFO [23968] Metadata: openslide.level[0].downsample = 1
INFO [23968] Metadata: openslide.level[0].height = 12288
INFO [23968] Metadata: openslide.level[0].tile-height = 128
INFO [23968] Metadata: openslide.level[0].tile-width = 128
INFO [23968] Metadata: openslide.level[0].width = 10240
INFO [23968] Metadata: openslide.level[1].downsample = 2
INFO [23968] Metadata: openslide.level[1].height = 6144
INFO [23968] Metadata: openslide.level[1].tile-height = 128
INFO [23968] Metadata: openslide.level[1].tile-width = 128
INFO [23968] Metadata: openslide.level[1].width = 5120
INFO [23968] Metadata: openslide.level[2].downsample = 4
INFO [23968] Metadata: openslide.level[2].height = 3072
INFO [23968] Metadata: openslide.level[2].tile-height = 128
INFO [23968] Metadata: openslide.level[2].tile-width = 128
INFO [23968] Metadata: openslide.level[2].width = 2560
INFO [23968] Metadata: openslide.level[3].downsample = 8
INFO [23968] Metadata: openslide.level[3].height = 1536
INFO [23968] Metadata: openslide.level[3].tile-height = 128
INFO [23968] Metadata: openslide.level[3].tile-width = 128
INFO [23968] Metadata: openslide.level[3].width = 1280
INFO [23968] Metadata: openslide.level[4].downsample = 16
INFO [23968] Metadata: openslide.level[4].height = 768
INFO [23968] Metadata: openslide.level[4].tile-height = 128
INFO [23968] Metadata: openslide.level[4].tile-width = 128
INFO [23968] Metadata: openslide.level[4].width = 640
INFO [23968] Metadata: openslide.level[5].downsample = 32
INFO [23968] Metadata: openslide.level[5].height = 384
INFO [23968] Metadata: openslide.level[5].tile-height = 128
INFO [23968] Metadata: openslide.level[5].tile-width = 128
INFO [23968] Metadata: openslide.level[5].width = 320
INFO [23968] Metadata: openslide.level[6].downsample = 64
INFO [23968] Metadata: openslide.level[6].height = 192
INFO [23968] Metadata: openslide.level[6].tile-height = 128
INFO [23968] Metadata: openslide.level[6].tile-width = 128
INFO [23968] Metadata: openslide.level[6].width = 160
INFO [23968] Metadata: openslide.level[7].downsample = 128
INFO [23968] Metadata: openslide.level[7].height = 96
INFO [23968] Metadata: openslide.level[7].tile-height = 128
INFO [23968] Metadata: openslide.level[7].tile-width = 128
INFO [23968] Metadata: openslide.level[7].width = 80
INFO [23968] Metadata: openslide.quickhash-1 = fd134bf84c0244baee8f25caeff5a7a977197905edb7064d0033f8f104c78941
INFO [23968] Metadata: openslide.vendor = generic-tiff
INFO [23968] Metadata: tiff.ImageDescription =
10240x12288 (256x256) JPEG/RGB Q=80|Date = 26/11/2021|Time = 15:56:12|MPP = 0.24984|3dh_PixelSizeX = 0.249711|3dh_PixelSizeY = 0.249968|3dh_Filter = Default|3dh_Profile = Routine x20|3dh_ScannerHardwareID = SC150-210905
INFO [23968] Metadata: tiff.ResolutionUnit = centimeter
INFO [23968] Metadata: tiff.XResolution = 40046.31640625
INFO [23968] Metadata: tiff.YResolution = 40005.28515625
INFO [23968] WSI has 8 levels
INFO [23968] Checking level 0
INFO [23968] WSI level 0 has size 10240x12288 and 480 MB adding..
INFO [23968] Setting tile width and height to 128 128
INFO [23968] Checking level 1
INFO [23968] WSI level 1 has size 5120x6144 and 120 MB adding..
INFO [23968] Setting tile width and height to 128 128
INFO [23968] Checking level 2
INFO [23968] WSI level 2 has size 2560x3072 and 30 MB adding..
INFO [23968] Setting tile width and height to 128 128
INFO [23968] Checking level 3
INFO [23968] WSI level 3 has size 1280x1536 and 7 MB adding..
INFO [23968] Setting tile width and height to 128 128
INFO [23968] Checking level 4
INFO [23968] WSI level was less than 4 MB, skipping..
count:0
calling insert renderer
INFO [23968] EXECUTING EmptyProcessObject because PO is modified.
INFO [23968] EXECUTING ImagePyramidRenderer because PO is modified.
INFO [23968] Recalculating the camera of the view
INFO [23968] Finished initializing OpenGL
finished insert renderer
INFO [23968] Recalculating the camera of the view
INFO [23968] Finished initializing OpenGL
INFO [23968] EXECUTING EmptyProcessObject because PO is modified.
INFO [23968] EXECUTING ImageChannelConverter because PO is modified.
Model name in wrapper: Epithelium_HE_UNET_512
Before update:
m[IE] = (TensorRT)
m[batch_process] = (2)
m[batch_size] = (1)
m[class_colors] = (0,0,255;0,255,0)
m[class_names] = (Exterior;Epithelium)
m[cpu] = (0)
m[input_img_size_x] = (512)
m[input_img_size_y] = (512)
m[input_node] = (ImageInputLayer)
m[interpolation] = (0)
m[magnification_level] = (10)
m[mask_threshold] = (0.02)
m[model_name] = (Epithelium_HE_512)
m[name] = (Epithelium_HE_512)
m[nb_channels] = (3)
m[nb_classes] = (2)
m[output_node] = (Softmax_Layer_Transpose2)
m[patch_overlap] = (0.05)
m[pipeline] = (import;tissue_segmentation;batchgen;neural_network;stitch;render)
m[problem] = (segmentation)
m[resolution] = (high)
m[scale_factor] = (1.0f/1.0f)
m[task] = (Epithelium_HE_512)
m[tissue_threshold] = (70)
IE:TensorRT
batch_process:2
batch_size:1
class_colors:0,0,255;0,255,0
class_names:Exterior;Epithelium
cpu:0
input_img_size_x:512
input_img_size_y:512
input_node:ImageInputLayer
interpolation:0
magnification_level:10
mask_threshold:0.02
model_name:Epithelium_HE_512
name:Epithelium_HE_512
nb_channels:3
nb_classes:2
output_node:Softmax_Layer_Transpose2
patch_overlap:0.05
pipeline:import;tissue_segmentation;batchgen;neural_network;stitch;render
problem:segmentation
resolution:high
scale_factor:1.0f/1.0f
task:Epithelium_HE_512
tissue_threshold:70
Ret value: 1
OK was pressed:
CANCEL was pressed:
Default...
After update:
m[IE] = (TensorRT)
m[batch_process] = (2)
m[batch_size] = (1)
m[class_colors] = (0,0,255;0,255,0)
m[class_names] = (Exterior;Epithelium)
m[cpu] = (0)
m[input_img_size_x] = (512)
m[input_img_size_y] = (512)
m[input_node] = (ImageInputLayer)
m[interpolation] = (0)
m[magnification_level] = (10)
m[mask_threshold] = (0.02)
m[model_name] = (Epithelium_HE_512)
m[name] = (Epithelium_HE_512)
m[nb_channels] = (3)
m[nb_classes] = (2)
m[output_node] = (Softmax_Layer_Transpose2)
m[patch_overlap] = (0.05)
m[pipeline] = (import;tissue_segmentation;batchgen;neural_network;stitch;render)
m[problem] = (segmentation)
m[resolution] = (high)
m[scale_factor] = (1.0f/1.0f)
m[task] = (Epithelium_HE_512)
m[tissue_threshold] = (70)
m[IE] = (TensorRT)
m[batch_process] = (2)
m[batch_size] = (1)
m[class_colors] = (0,0,255;0,255,0)
m[class_names] = (Exterior;Epithelium)
m[cpu] = (0)
m[input_img_size_x] = (512)
m[input_img_size_y] = (512)
m[input_node] = (ImageInputLayer)
m[interpolation] = (0)
m[magnification_level] = (10)
m[mask_threshold] = (0.02)
m[model_name] = (Epithelium_HE_512)
m[name] = (Epithelium_HE_512)
m[nb_channels] = (3)
m[nb_classes] = (2)
m[output_node] = (Softmax_Layer_Transpose2)
m[patch_overlap] = (0.05)
m[pipeline] = (import;tissue_segmentation;batchgen;neural_network;stitch;render)
m[problem] = (segmentation)
m[resolution] = (high)
m[scale_factor] = (1.0f/1.0f)
m[task] = (Epithelium_HE_512)
m[tissue_threshold] = (70)
Final model metadata config sent to pixelClassifier:
m[IE] = (TensorRT)
m[batch_process] = (2)
m[batch_size] = (1)
m[class_colors] = (0,0,255;0,255,0)
m[class_names] = (Exterior;Epithelium)
m[cpu] = (0)
m[input_img_size_x] = (512)
m[input_img_size_y] = (512)
m[input_node] = (ImageInputLayer)
m[interpolation] = (0)
m[magnification_level] = (10)
m[mask_threshold] = (0.02)
m[model_name] = (Epithelium_HE_512)
m[name] = (Epithelium_HE_512)
m[nb_channels] = (3)
m[nb_classes] = (2)
m[output_node] = (Softmax_Layer_Transpose2)
m[patch_overlap] = (0.05)
m[pipeline] = (import;tissue_segmentation;batchgen;neural_network;stitch;render)
m[problem] = (segmentation)
m[resolution] = (high)
m[scale_factor] = (1.0f/1.0f)
m[task] = (Epithelium_HE_512)
m[tissue_threshold] = (70)
Final model metadata config WITHIN to pixelClassifier:
m[IE] = (TensorRT)
m[batch_process] = (2)
m[batch_size] = (1)
m[class_colors] = (0,0,255;0,255,0)
m[class_names] = (Exterior;Epithelium)
m[cpu] = (0)
m[input_img_size_x] = (512)
m[input_img_size_y] = (512)
m[input_node] = (ImageInputLayer)
m[interpolation] = (0)
m[magnification_level] = (10)
m[mask_threshold] = (0.02)
m[model_name] = (Epithelium_HE_512)
m[name] = (Epithelium_HE_512)
m[nb_channels] = (3)
m[nb_classes] = (2)
m[output_node] = (Softmax_Layer_Transpose2)
m[patch_overlap] = (0.05)
m[pipeline] = (import;tissue_segmentation;batchgen;neural_network;stitch;render)
m[problem] = (segmentation)
m[resolution] = (high)
m[scale_factor] = (1.0f/1.0f)
m[task] = (Epithelium_HE_512)
m[tissue_threshold] = (70)
Current model: Epithelium_HE_UNET_512
current WSI: D:/Pradeep/Andre/pyramidal_conv/pyramidal.tif
Curr patch level: 2
current WSI: D:/Pradeep/Andre/pyramidal_conv/pyramidal.tif
INFO [23968] EXECUTING WholeSlideImageImporter because PO is modified.
INFO [23968] Metadata: openslide.comment =
10240x12288 (256x256) JPEG/RGB Q=80|Date = 26/11/2021|Time = 15:56:12|MPP = 0.24984|3dh_PixelSizeX = 0.249711|3dh_PixelSizeY = 0.249968|3dh_Filter = Default|3dh_Profile = Routine x20|3dh_ScannerHardwareID = SC150-210905
INFO [23968] Metadata: openslide.level-count = 8
INFO [23968] Metadata: openslide.level[0].downsample = 1
INFO [23968] Metadata: openslide.level[0].height = 12288
INFO [23968] Metadata: openslide.level[0].tile-height = 128
INFO [23968] Metadata: openslide.level[0].tile-width = 128
INFO [23968] Metadata: openslide.level[0].width = 10240
INFO [23968] Metadata: openslide.level[1].downsample = 2
INFO [23968] Metadata: openslide.level[1].height = 6144
INFO [23968] Metadata: openslide.level[1].tile-height = 128
INFO [23968] Metadata: openslide.level[1].tile-width = 128
INFO [23968] Metadata: openslide.level[1].width = 5120
INFO [23968] Metadata: openslide.level[2].downsample = 4
INFO [23968] Metadata: openslide.level[2].height = 3072
INFO [23968] Metadata: openslide.level[2].tile-height = 128
INFO [23968] Metadata: openslide.level[2].tile-width = 128
INFO [23968] Metadata: openslide.level[2].width = 2560
INFO [23968] Metadata: openslide.level[3].downsample = 8
INFO [23968] Metadata: openslide.level[3].height = 1536
INFO [23968] Metadata: openslide.level[3].tile-height = 128
INFO [23968] Metadata: openslide.level[3].tile-width = 128
INFO [23968] Metadata: openslide.level[3].width = 1280
INFO [23968] Metadata: openslide.level[4].downsample = 16
INFO [23968] Metadata: openslide.level[4].height = 768
INFO [23968] Metadata: openslide.level[4].tile-height = 128
INFO [23968] Metadata: openslide.level[4].tile-width = 128
INFO [23968] Metadata: openslide.level[4].width = 640
INFO [23968] Metadata: openslide.level[5].downsample = 32
INFO [23968] Metadata: openslide.level[5].height = 384
INFO [23968] Metadata: openslide.level[5].tile-height = 128
INFO [23968] Metadata: openslide.level[5].tile-width = 128
INFO [23968] Metadata: openslide.level[5].width = 320
INFO [23968] Metadata: openslide.level[6].downsample = 64
INFO [23968] Metadata: openslide.level[6].height = 192
INFO [23968] Metadata: openslide.level[6].tile-height = 128
INFO [23968] Metadata: openslide.level[6].tile-width = 128
INFO [23968] Metadata: openslide.level[6].width = 160
INFO [23968] Metadata: openslide.level[7].downsample = 128
INFO [23968] Metadata: openslide.level[7].height = 96
INFO [23968] Metadata: openslide.level[7].tile-height = 128
INFO [23968] Metadata: openslide.level[7].tile-width = 128
INFO [23968] Metadata: openslide.level[7].width = 80
INFO [23968] Metadata: openslide.quickhash-1 = fd134bf84c0244baee8f25caeff5a7a977197905edb7064d0033f8f104c78941
INFO [23968] Metadata: openslide.vendor = generic-tiff
INFO [23968] Metadata: tiff.ImageDescription =
10240x12288 (256x256) JPEG/RGB Q=80|Date = 26/11/2021|Time = 15:56:12|MPP = 0.24984|3dh_PixelSizeX = 0.249711|3dh_PixelSizeY = 0.249968|3dh_Filter = Default|3dh_Profile = Routine x20|3dh_ScannerHardwareID = SC150-210905
INFO [23968] Metadata: tiff.ResolutionUnit = centimeter
INFO [23968] Metadata: tiff.XResolution = 40046.31640625
INFO [23968] Metadata: tiff.YResolution = 40005.28515625
INFO [23968] WSI has 8 levels
INFO [23968] Checking level 0
INFO [23968] WSI level 0 has size 10240x12288 and 480 MB adding..
INFO [23968] Setting tile width and height to 128 128
INFO [23968] Checking level 1
INFO [23968] WSI level 1 has size 5120x6144 and 120 MB adding..
INFO [23968] Setting tile width and height to 128 128
INFO [23968] Checking level 2
INFO [23968] WSI level 2 has size 2560x3072 and 30 MB adding..
INFO [23968] Setting tile width and height to 128 128
INFO [23968] Checking level 3
INFO [23968] WSI level 3 has size 1280x1536 and 7 MB adding..
INFO [23968] Setting tile width and height to 128 128
INFO [23968] Checking level 4
INFO [23968] WSI level was less than 4 MB, skipping..
Current OS is: windows
Current kernel is: winnt
accepted models: .onnx
accepted models: .txt
INFO [23968] Loading inference engines in folder C:/Program Files/FastPathology/bin//..//bin/
INFO [23968] Loading inference engine OpenVINO from shared library InferenceEngineOpenVINO.dll
INFO [23968] Loading inference engine TensorFlow from shared library InferenceEngineTensorFlow.dll
2021-12-10 18:35:26.944982: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library cudart64_110.dll
INFO [23968] Loading inference engine TensorRT from shared library InferenceEngineTensorRT.dll
INFO [23968] Inference engine TensorRT selected
INFO [23968] Inference engine TensorRT selected
Current available IEs:
OpenVINO,
TensorFlow,
TensorRT,
Which model formats are available and that there exists an IE for:
.onnx,
.txt,
TensorRT (using ONNX) selected
INFO [23968] Inference engine TensorRT selected
Model was found.
Preselected IE was used: TensorRT
INFO [23968] Inference engine TensorRT selected
INFO [23968] Serialized file C:/ProgramData/FAST/kernel_binaries///Epithelium_HE_UNET_512.onnx_653401940713773443.bin is up to date.
INFO [23968] [MemUsageChange] Init CUDA: CPU +263, GPU +0, now: CPU 14965, GPU 1146 (MiB)
INFO [23968] Loaded engine size: 1245 MiB
WARNING [23968] TensorRT was linked against cuBLAS/cuBLAS LT 11.6.3 but loaded cuBLAS/cuBLAS LT 11.4.1
INFO [23968] [MemUsageChange] Init cuBLAS/cuBLASLt: CPU +216, GPU +76, now: CPU 16349, GPU 2468 (MiB)
INFO [23968] [MemUsageChange] Init cuDNN: CPU +250, GPU +78, now: CPU 16599, GPU 2546 (MiB)
INFO [23968] [MemUsageChange] TensorRT-managed allocation in engine deserialization: CPU +0, GPU +1245, now: CPU 0, GPU 1245 (MiB)
INFO [23968] Guessed image ordering to be channel first as shape was 1 3 512 512
INFO [23968] Found input node ImageInputLayer with shape 1 3 512 512
INFO [23968] Guessed image ordering to be channel first as shape was 1 2 512 512
INFO [23968] Found output node Softmax_Layer_Transpose2 with shape 1 2 512 512
WARNING [23968] TensorRT was linked against cuBLAS/cuBLAS LT 11.6.3 but loaded cuBLAS/cuBLAS LT 11.4.1
INFO [23968] [MemUsageChange] Init cuBLAS/cuBLASLt: CPU +0, GPU +10, now: CPU 15354, GPU 2538 (MiB)
INFO [23968] [MemUsageChange] Init cuDNN: CPU +0, GPU +8, now: CPU 15354, GPU 2546 (MiB)
INFO [23968] [MemUsageChange] TensorRT-managed allocation in IExecutionContext creation: CPU +0, GPU +1362, now: CPU 0, GPU 2607 (MiB)
Threshold was defined: 70
tissue_threshold was defined, so is performing thresholding as preprocessing step.
Setting mask_threshold to: 0.02
calling insert renderer
INFO [23968] EXECUTING EmptyProcessObject because PO is modified.
INFO [23968] EXECUTING TissueSegmentation because PO is modified.
INFO [23968] EXECUTING EmptyProcessObject because PO is modified.
INFO [23968] EXECUTING ImageChannelConverter because PO is modified.
INFO [23968] EXECUTING EmptyProcessObject because PO is modified.
INFO [23968] EXECUTING Dilation because PO is modified.
INFO [23968] EXECUTING Erosion because PO is modified.
INFO [23968] EXECUTING EmptyProcessObject because PO is modified.
INFO [23968] EXECUTING PatchGenerator because PO is modified.
WARNING [7648] Patch size must be a multiple of 16 (TIFF limitation). Adding some overlap (32, 32) to fix.
INFO [7648] Generating patch 2 1
INFO [7648] EXECUTING EmptyProcessObject because PO is modified.
INFO [7648] EXECUTING ImageChannelConverter because PO is modified.
INFO [7648] Generating patch 3 1
INFO [7648] EXECUTING EmptyProcessObject because PO is modified.
INFO [7648] EXECUTING ImageChannelConverter because PO is modified.
INFO [23968] EXECUTING SegmentationNetwork because PO is modified.
INFO [7648] Generating patch 1 2
INFO [23968] Finished copying input data to TensorRT
INFO [7648] EXECUTING EmptyProcessObject because PO is modified.
INFO [7648] EXECUTING ImageChannelConverter because PO is modified.
INFO [7648] Generating patch 2 2
INFO [23968] Finished execute TensorRT
INFO [23968] Finished freeing input data TensorRT
INFO [23968] Processing output node Softmax_Layer_Transpose2INFO [7648] EXECUTING EmptyProcessObject because PO is modified.

INFO [23968] Finished moving data to FAST tensor, TensorRTINFO [7648] EXECUTING ImageChannelConverter because PO is modified.
INFO [23968]
Finished transfer of output data TensorRT
INFO [23968] Finished freeing output data TensorRTINFO [7648] Generating patch 3 2

INFO [23968] EXECUTING EmptyProcessObject because PO is modified.
INFO [23968] EXECUTING TensorToSegmentation because PO is modified.
INFO [23968] EXECUTING PatchStitcher because PO is modified.
INFO [23968] Patch stitcher creating image with size 2560 3072
INFO [23968] INFO [7648] EXECUTING EmptyProcessObject because PO is modified.Stitching 2D data 2 1
INFO [7648] EXECUTING ImageChannelConverter because PO is modified.

INFO [23968] EXECUTING SegmentationRenderer because PO is modified.INFO [7648] Generating patch 4 2

INFO [23968] Recalculating the camera of the view
INFO [23968] Finished initializing OpenGL
finished insert renderer
Current save location: C:/Users/rajasekhar.p/AppData/Local/Temp/fastpathology-MlbLYw/project_17409488
INFO [15736] EXECUTING SegmentationNetwork because PO has new input data.INFO [7648] EXECUTING EmptyProcessObject because PO is modified.

INFO [7648] EXECUTING ImageChannelConverter because PO is modified.
INFO [7648] Generating patch 5 2
INFO [15736] Finished copying input data to TensorRT
INFO [7648] EXECUTING EmptyProcessObject because PO is modified.
INFO [7648] EXECUTING ImageChannelConverter because PO is modified.
INFO [7648] Generating patch 1 3
INFO [15736] Finished execute TensorRTInference thread is finished...

INFO [15736] Finished freeing input data TensorRT
INFO [15736] Processing output node Softmax_Layer_Transpose2
INFO [15736] Finished moving data to FAST tensor, TensorRT
INFO [15736] Finished transfer of output data TensorRT
INFO [15736] Finished freeing output data TensorRTINFO [7648] EXECUTING EmptyProcessObject because PO is modified.

INFO [7648] EXECUTING ImageChannelConverter because PO is modified.INFO [
15736] EXECUTING EmptyProcessObject because PO is modified.INFO [7648] Generating patch 2 3
INFO [15736]
EXECUTING TensorToSegmentation because PO is modified.
INFO [15736] EXECUTING PatchStitcher because PO has new input data.
INFO [15736] Stitching 2D data 3 1
INFO [15736] EXECUTING SegmentationNetwork because PO has new input data.
INFO [15736] Finished copying input data to TensorRT
INFO [7648] EXECUTING EmptyProcessObject because PO is modified.
INFO [7648] EXECUTING ImageChannelConverter because PO is modified.
INFO [7648] Generating patch 3 3
INFO [15736] Finished execute TensorRT
INFO [15736] Finished freeing input data TensorRT
INFO [15736] Processing output node Softmax_Layer_Transpose2
INFO [7648] EXECUTING EmptyProcessObject because PO is modified.INFO [15736] Finished moving data to FAST tensor, TensorRT
INFO [7648] EXECUTING ImageChannelConverter because PO is modified.
INFO [15736] Finished transfer of output data TensorRT

INFO [15736] Finished freeing output data TensorRT
INFO [7648] Generating patch 4 3
INFO [15736] EXECUTING EmptyProcessObject because PO is modified.
INFO [15736] EXECUTING TensorToSegmentation because PO is modified.
INFO [15736] EXECUTING PatchStitcher because PO has new input data.
INFO [15736] Stitching 2D data 1 2
INFO [15736] EXECUTING SegmentationNetwork because PO has new input data.
INFO [15736] Finished copying input data to TensorRT
INFO [7648] EXECUTING EmptyProcessObject because PO is modified.
INFO [7648] EXECUTING ImageChannelConverter because PO is modified.
INFO [7648] Generating patch 5 3
INFO [7648] EXECUTING EmptyProcessObject because PO is modified.
INFO [7648] EXECUTING ImageChannelConverter because PO is modified.
INFO [15736] Finished execute TensorRTINFO [7648] Generating patch 1 4

INFO [15736] Finished freeing input data TensorRT
INFO [15736] Processing output node Softmax_Layer_Transpose2
INFO [15736] Finished moving data to FAST tensor, TensorRT
INFO [15736] Finished transfer of output data TensorRT
INFO [15736] Finished freeing output data TensorRT
INFO [15736] EXECUTING EmptyProcessObject because PO is modified.
INFO [15736] EXECUTING TensorToSegmentation because PO is modified.
INFO [15736] EXECUTING PatchStitcher because PO has new input data.
INFO [15736] Stitching 2D data 2 2
INFO [15736] EXECUTING SegmentationNetwork because PO has new input data.
INFO [7648] EXECUTING EmptyProcessObject because PO is modified.INFO [15736] Finished copying input data to TensorRT
INFO [7648] EXECUTING ImageChannelConverter because PO is modified.

INFO [7648] Generating patch 2 4
INFO [7648] EXECUTING EmptyProcessObject because PO is modified.
INFO [7648] EXECUTING ImageChannelConverter because PO is modified.
INFO [15736] Finished execute TensorRTINFO [7648] Generating patch 3 4

INFO [15736] Finished freeing input data TensorRT
INFO [15736] Processing output node Softmax_Layer_Transpose2
INFO [15736] Finished moving data to FAST tensor, TensorRT
INFO [15736] Finished transfer of output data TensorRT
INFO [15736] Finished freeing output data TensorRT
INFO [15736] EXECUTING EmptyProcessObject because PO is modified.
INFO [15736] EXECUTING TensorToSegmentation because PO is modified.
INFO [15736] EXECUTING PatchStitcher because PO has new input data.
INFO [15736] Stitching 2D data 3 2
INFO [15736] EXECUTING SegmentationNetwork because PO has new input data.
INFO [15736] Finished copying input data to TensorRT
INFO [7648] EXECUTING EmptyProcessObject because PO is modified.
INFO [7648] EXECUTING ImageChannelConverter because PO is modified.
INFO [7648] Generating patch 4 4
INFO [15736] Finished execute TensorRT
INFO [15736] Finished freeing input data TensorRT
INFO [15736] Processing output node Softmax_Layer_Transpose2
INFO [7648] EXECUTING EmptyProcessObject because PO is modified.INFO [15736] Finished moving data to FAST tensor, TensorRT
INFO [7648] EXECUTING ImageChannelConverter because PO is modified.
INFO [15736] Finished transfer of output data TensorRT

INFO [15736] Finished freeing output data TensorRTINFO [7648] Generating patch 5 4

INFO [15736] EXECUTING EmptyProcessObject because PO is modified.
INFO [15736] EXECUTING TensorToSegmentation because PO is modified.
INFO [15736] EXECUTING PatchStitcher because PO has new input data.
INFO [15736] Stitching 2D data 4 2
INFO [15736] EXECUTING SegmentationNetwork because PO has new input data.
INFO [15736] EXECUTING EmptyProcessObject because PO is modified.
INFO [15736] EXECUTING ImageResizer because PO is modified.INFO [7648] EXECUTING EmptyProcessObject because PO is modified.

INFO [7648] EXECUTING ImageChannelConverter because PO is modified.
INFO [7648] Generating patch 1 5
INFO [15736] Finished copying input data to TensorRT
INFO [7648] EXECUTING EmptyProcessObject because PO is modified.
INFO [7648] EXECUTING ImageChannelConverter because PO is modified.
INFO [7648] Generating patch 2 5
INFO [15736] Finished execute TensorRT
INFO [15736] Finished freeing input data TensorRT
INFO [15736] Processing output node Softmax_Layer_Transpose2
INFO [15736] Finished moving data to FAST tensor, TensorRT
INFO [15736] Finished transfer of output data TensorRT
INFO [15736] Finished freeing output data TensorRT
INFO [15736] EXECUTING EmptyProcessObject because PO is modified.
INFO [15736] EXECUTING TensorToSegmentation because PO is modified.
INFO [15736] EXECUTING PatchStitcher because PO has new input data.
INFO [15736] Stitching 2D data 5 2
INFO [7648] EXECUTING EmptyProcessObject because PO is modified.INFO [15736] EXECUTING SegmentationNetwork because PO has new input data.
INFO [7648] EXECUTING ImageChannelConverter because PO is modified.

INFO [7648] Generating patch 3 5
INFO [15736] Finished copying input data to TensorRT
INFO [7648] EXECUTING EmptyProcessObject because PO is modified.
INFO [7648] EXECUTING ImageChannelConverter because PO is modified.
INFO [7648] Generating patch 4 5
INFO [15736] Finished execute TensorRT
INFO [15736] Finished freeing input data TensorRT
INFO [15736] Processing output node Softmax_Layer_Transpose2
INFO [15736] Finished moving data to FAST tensor, TensorRT
INFO [15736] Finished transfer of output data TensorRT
INFO [15736] Finished freeing output data TensorRT
INFO [15736] EXECUTING EmptyProcessObject because PO is modified.
INFO [15736] EXECUTING TensorToSegmentation because PO is modified.INFO [7648] EXECUTING EmptyProcessObject because PO is modified.

INFO [7648] EXECUTING ImageChannelConverter because PO is modified.INFO [15736] EXECUTING PatchStitcher because PO has new input data.

INFO [15736] Stitching 2D data 1 3INFO [7648] Generating patch 5 5

INFO [15736] EXECUTING SegmentationNetwork because PO has new input data.
INFO [15736] Finished copying input data to TensorRT
INFO [7648] EXECUTING EmptyProcessObject because PO is modified.
INFO [7648] EXECUTING ImageChannelConverter because PO is modified.
INFO [7648] Generating patch 2 6
INFO [15736] Finished execute TensorRT
INFO [15736] Finished freeing input data TensorRT
INFO [15736] Processing output node Softmax_Layer_Transpose2
INFO [15736] Finished moving data to FAST tensor, TensorRT
INFO [15736] Finished transfer of output data TensorRT
INFO [15736] Finished freeing output data TensorRT
INFO [15736] EXECUTING EmptyProcessObject because PO is modified.
INFO [15736] EXECUTING TensorToSegmentation because PO is modified.
INFO [15736] EXECUTING PatchStitcher because PO has new input data.
INFO [15736] Stitching 2D data 2 3
INFO [15736] EXECUTING SegmentationNetwork because PO has new input data.
INFO [15736] Finished copying input data to TensorRT
INFO [7648] EXECUTING EmptyProcessObject because PO is modified.
INFO [7648] EXECUTING ImageChannelConverter because PO is modified.
INFO [7648] Generating patch 3 6
INFO [7648] EXECUTING EmptyProcessObject because PO is modified.
INFO [7648] EXECUTING ImageChannelConverter because PO is modified.INFO [15736] Finished execute TensorRT

INFO [15736] Finished freeing input data TensorRTINFO [7648] Done generating patches
INFO [15736] Processing output node Softmax_Layer_Transpose2

INFO [15736] Finished moving data to FAST tensor, TensorRT
INFO [15736] Finished transfer of output data TensorRT
INFO [15736] Finished freeing output data TensorRT
INFO [15736] EXECUTING EmptyProcessObject because PO is modified.
INFO [15736] EXECUTING TensorToSegmentation because PO is modified.
INFO [15736] EXECUTING PatchStitcher because PO has new input data.
INFO [15736] Stitching 2D data 3 3
INFO [15736] EXECUTING SegmentationNetwork because PO has new input data.
INFO [15736] Finished copying input data to TensorRT
INFO [15736] Finished execute TensorRT
INFO [15736] Finished freeing input data TensorRT
INFO [15736] Processing output node Softmax_Layer_Transpose2
INFO [15736] Finished moving data to FAST tensor, TensorRT
INFO [15736] Finished transfer of output data TensorRT
INFO [15736] Finished freeing output data TensorRT
INFO [15736] EXECUTING EmptyProcessObject because PO is modified.
INFO [15736] EXECUTING TensorToSegmentation because PO is modified.
INFO [15736] EXECUTING PatchStitcher because PO has new input data.
INFO [15736] Stitching 2D data 4 3
INFO [15736] EXECUTING SegmentationNetwork because PO has new input data.
INFO [15736] EXECUTING EmptyProcessObject because PO is modified.
INFO [15736] EXECUTING ImageResizer because PO is modified.
INFO [15736] Finished copying input data to TensorRT
INFO [15736] Finished execute TensorRT
INFO [15736] Finished freeing input data TensorRT
INFO [15736] Processing output node Softmax_Layer_Transpose2
INFO [15736] Finished moving data to FAST tensor, TensorRT
INFO [15736] Finished transfer of output data TensorRT
INFO [15736] Finished freeing output data TensorRT
INFO [15736] EXECUTING EmptyProcessObject because PO is modified.
INFO [15736] EXECUTING TensorToSegmentation because PO is modified.
INFO [15736] EXECUTING PatchStitcher because PO has new input data.
INFO [15736] Stitching 2D data 5 3
INFO [15736] EXECUTING SegmentationNetwork because PO has new input data.
INFO [15736] Finished copying input data to TensorRT
INFO [15736] Finished execute TensorRT
INFO [15736] Finished freeing input data TensorRT
INFO [15736] Processing output node Softmax_Layer_Transpose2
INFO [15736] Finished moving data to FAST tensor, TensorRT
INFO [15736] Finished transfer of output data TensorRT
INFO [15736] Finished freeing output data TensorRT
INFO [15736] EXECUTING EmptyProcessObject because PO is modified.
INFO [15736] EXECUTING TensorToSegmentation because PO is modified.
INFO [15736] EXECUTING PatchStitcher because PO has new input data.
INFO [15736] Stitching 2D data 1 4
INFO [15736] EXECUTING SegmentationNetwork because PO has new input data.
INFO [15736] Finished copying input data to TensorRT
INFO [15736] Finished execute TensorRT
INFO [15736] Finished freeing input data TensorRT
INFO [15736] Processing output node Softmax_Layer_Transpose2
INFO [15736] Finished moving data to FAST tensor, TensorRT
INFO [15736] Finished transfer of output data TensorRT
INFO [15736] Finished freeing output data TensorRT
INFO [15736] EXECUTING EmptyProcessObject because PO is modified.
INFO [15736] EXECUTING TensorToSegmentation because PO is modified.
INFO [15736] EXECUTING PatchStitcher because PO has new input data.
INFO [15736] Stitching 2D data 2 4
INFO [15736] EXECUTING SegmentationNetwork because PO has new input data.
INFO [15736] Finished copying input data to TensorRT
INFO [15736] Finished execute TensorRT
INFO [15736] Finished freeing input data TensorRT
INFO [15736] Processing output node Softmax_Layer_Transpose2
INFO [15736] Finished moving data to FAST tensor, TensorRT
INFO [15736] Finished transfer of output data TensorRT
INFO [15736] Finished freeing output data TensorRT
INFO [15736] EXECUTING EmptyProcessObject because PO is modified.
INFO [15736] EXECUTING TensorToSegmentation because PO is modified.
INFO [15736] EXECUTING PatchStitcher because PO has new input data.
INFO [15736] Stitching 2D data 3 4
INFO [15736] EXECUTING SegmentationNetwork because PO has new input data.
INFO [15736] Finished copying input data to TensorRT
INFO [15736] Finished execute TensorRT
INFO [15736] Finished freeing input data TensorRT
INFO [15736] Processing output node Softmax_Layer_Transpose2
INFO [15736] Finished moving data to FAST tensor, TensorRT
INFO [15736] Finished transfer of output data TensorRT
INFO [15736] Finished freeing output data TensorRT
INFO [15736] EXECUTING EmptyProcessObject because PO is modified.
INFO [15736] EXECUTING TensorToSegmentation because PO is modified.
INFO [15736] EXECUTING PatchStitcher because PO has new input data.
INFO [15736] Stitching 2D data 4 4
INFO [15736] EXECUTING SegmentationNetwork because PO has new input data.
INFO [15736] EXECUTING EmptyProcessObject because PO is modified.
INFO [15736] EXECUTING ImageResizer because PO is modified.
INFO [15736] Finished copying input data to TensorRT
INFO [15736] Finished execute TensorRT
INFO [15736] Finished freeing input data TensorRT
INFO [15736] Processing output node Softmax_Layer_Transpose2
INFO [15736] Finished moving data to FAST tensor, TensorRT
INFO [15736] Finished transfer of output data TensorRT
INFO [15736] Finished freeing output data TensorRT
INFO [15736] EXECUTING EmptyProcessObject because PO is modified.
INFO [15736] EXECUTING TensorToSegmentation because PO is modified.
INFO [15736] EXECUTING PatchStitcher because PO has new input data.
INFO [15736] Stitching 2D data 5 4
INFO [15736] EXECUTING SegmentationNetwork because PO has new input data.
INFO [15736] Finished copying input data to TensorRT
INFO [15736] Finished execute TensorRT
INFO [15736] Finished freeing input data TensorRT
INFO [15736] Processing output node Softmax_Layer_Transpose2
INFO [15736] Finished moving data to FAST tensor, TensorRT
INFO [15736] Finished transfer of output data TensorRT
INFO [15736] Finished freeing output data TensorRT
INFO [15736] EXECUTING EmptyProcessObject because PO is modified.
INFO [15736] EXECUTING TensorToSegmentation because PO is modified.
INFO [15736] EXECUTING PatchStitcher because PO has new input data.
INFO [15736] Stitching 2D data 1 5
INFO [15736] EXECUTING SegmentationNetwork because PO has new input data.
INFO [15736] Finished copying input data to TensorRT
INFO [15736] Finished execute TensorRT
INFO [15736] Finished freeing input data TensorRT
INFO [15736] Processing output node Softmax_Layer_Transpose2
INFO [15736] Finished moving data to FAST tensor, TensorRT
INFO [15736] Finished transfer of output data TensorRT
INFO [15736] Finished freeing output data TensorRT
INFO [15736] EXECUTING EmptyProcessObject because PO is modified.
INFO [15736] EXECUTING TensorToSegmentation because PO is modified.
INFO [15736] EXECUTING PatchStitcher because PO has new input data.
INFO [15736] Stitching 2D data 2 5
INFO [15736] EXECUTING SegmentationNetwork because PO has new input data.
INFO [15736] Finished copying input data to TensorRT
INFO [15736] Finished execute TensorRT
INFO [15736] Finished freeing input data TensorRT
INFO [15736] Processing output node Softmax_Layer_Transpose2
INFO [15736] Finished moving data to FAST tensor, TensorRT
INFO [15736] Finished transfer of output data TensorRT
INFO [15736] Finished freeing output data TensorRT
INFO [15736] EXECUTING EmptyProcessObject because PO is modified.
INFO [15736] EXECUTING TensorToSegmentation because PO is modified.
INFO [15736] EXECUTING PatchStitcher because PO has new input data.
INFO [15736] Stitching 2D data 3 5
INFO [15736] EXECUTING SegmentationNetwork because PO has new input data.
INFO [15736] Finished copying input data to TensorRT
INFO [15736] Finished execute TensorRT
INFO [15736] Finished freeing input data TensorRT
INFO [15736] Processing output node Softmax_Layer_Transpose2
INFO [15736] Finished moving data to FAST tensor, TensorRT
INFO [15736] Finished transfer of output data TensorRT
INFO [15736] Finished freeing output data TensorRT
INFO [15736] EXECUTING EmptyProcessObject because PO is modified.
INFO [15736] EXECUTING TensorToSegmentation because PO is modified.
INFO [15736] EXECUTING PatchStitcher because PO has new input data.
INFO [15736] Stitching 2D data 4 5
INFO [15736] EXECUTING SegmentationNetwork because PO has new input data.
INFO [15736] EXECUTING EmptyProcessObject because PO is modified.
INFO [15736] EXECUTING ImageResizer because PO is modified.
INFO [15736] Finished copying input data to TensorRT
INFO [15736] Finished execute TensorRT
INFO [15736] Finished freeing input data TensorRT
INFO [15736] Processing output node Softmax_Layer_Transpose2
INFO [15736] Finished moving data to FAST tensor, TensorRT
INFO [15736] Finished transfer of output data TensorRT
INFO [15736] Finished freeing output data TensorRT
INFO [15736] EXECUTING EmptyProcessObject because PO is modified.
INFO [15736] EXECUTING TensorToSegmentation because PO is modified.
INFO [15736] EXECUTING PatchStitcher because PO has new input data.
INFO [15736] Stitching 2D data 5 5
INFO [15736] EXECUTING SegmentationNetwork because PO has new input data.
INFO [15736] EXECUTING EmptyProcessObject because PO is modified.
INFO [15736] EXECUTING ImageResizer because PO is modified.
INFO [15736] Finished copying input data to TensorRT
INFO [15736] Finished execute TensorRT
INFO [15736] Finished freeing input data TensorRT
INFO [15736] Processing output node Softmax_Layer_Transpose2
INFO [15736] Finished moving data to FAST tensor, TensorRT
INFO [15736] Finished transfer of output data TensorRT
INFO [15736] Finished freeing output data TensorRT
INFO [15736] EXECUTING EmptyProcessObject because PO is modified.
INFO [15736] EXECUTING TensorToSegmentation because PO is modified.
INFO [15736] EXECUTING PatchStitcher because PO has new input data.
INFO [15736] Stitching 2D data 2 6
INFO [15736] EXECUTING SegmentationNetwork because PO has new input data.
INFO [15736] EXECUTING EmptyProcessObject because PO is modified.
INFO [15736] EXECUTING ImageResizer because PO is modified.
INFO [15736] Finished copying input data to TensorRT
INFO [15736] Finished execute TensorRT
INFO [15736] Finished freeing input data TensorRT
INFO [15736] Processing output node Softmax_Layer_Transpose2
INFO [15736] Finished moving data to FAST tensor, TensorRT
INFO [15736] Finished transfer of output data TensorRT
INFO [15736] Finished freeing output data TensorRT
INFO [15736] EXECUTING EmptyProcessObject because PO is modified.
INFO [15736] EXECUTING TensorToSegmentation because PO is modified.
INFO [15736] EXECUTING PatchStitcher because PO has new input data.
INFO [15736] Stitching 2D data 3 6
INFO [15736] EXECUTING TIFFImagePyramidExporter because PO is modified.
INFO [15736] Data given to TIFFImagePyramidExporter was an Image, not an ImagePyramid, converting ...

[andreped]

I didn't realise I had to install TensorRT. I've done that and changed the model config file so its "TensorRT"
I also have CUDA and cudnn installed.

To run inference using the GPU and TensorRT, you need TensorRT installed. However, you do not need TensorRT to run inference. You can run inference using for instance OpenVINO (which uses the CPU or the integrated GPU - if you have one). TensorRT uses a dedicated GPU (e.g. NVIDIA).

Strange that it is crashing at the end. Seems like it crashes when trying to export, that is save the result on disk. In order to save something on disk, it requires that this path has been set. When the program is initialized, there is created a temporary folder, where there is created a temporary Project folder, where results are to be saved to. Perhaps the directory or Project folder was not properly initialized.

Could you try creating a Project beforehand, and then running inference? It is similar to how you would do it in QuPath. I recommend creating the Project and then adding the images to the Project, as you normally would.

[pr4deepr]

I ran into the same issue again.
I created a project and ran the inference, which crashed..
I reopened the project and ran it again. Still had the same issue..

INFO [11340] Loaded configuration file: C:/Program Files/FastPathology/bin/fast_configuration.txt
INFO [11340] Test data path: C:/ProgramData/FAST/data/
INFO [11340] Kernel source path: C:/Program Files/FastPathology/bin//..//kernels/
INFO [11340] Kernel binary path: C:/ProgramData/FAST/kernel_binaries/
INFO [11340] Documentation path: C:/Program Files/FastPathology/bin//..//doc/
INFO [11340] Pipeline path: C:/ProgramData/FAST/pipelines/
INFO [11340] Qt plugins path: C:/Program Files/FastPathology/bin//..//plugins/
INFO [11340] Library path: C:/Program Files/FastPathology/bin//..//bin/
INFO [11340] Creating new QApp
INFO [11340] Creating new GL context for computation thread
INFO [11340] Medium large screen detected with width: 2560
Temporary path (UPDATED!): C:/Users/rajasekhar.p/AppData/Local/Temp/fastpathology-mXlwLC
Does folder exist:1
Current temporary project folder location:C:/Users/rajasekhar.p/AppData/Local/Temp/fastpathology-mXlwLC/project_17409488
INFO [11340] QApp already exists..
INFO [11340] Found 1 OpenCL platforms.
INFO [11340] 1 platforms selected for inspection.
INFO [11340] Platform 0: NVIDIA Corporation
INFO [11340] This platform has 1 available devices in total
INFO [11340] Looking for GPU devices only.
INFO [11340] 1 selected.
INFO [11340] Inspecting device 0 with the name Quadro P4000
INFO [11340] There are 1 devices that can be associated with the GL context
INFO [11340] 000002733ADC1D70 - 000002733ADC1D70
INFO [11340] Device has OpenGL interop capability
INFO [11340] The device was accepted.
INFO [11340] Quadro P4000 has 14 compute units
INFO [11340] The device Quadro P4000 got a score of 1000014
INFO [11340] The platform NVIDIA CUDA was selected as the best platform.
INFO [11340] A total of 1 devices were selected for the context from this platform:
INFO [11340] The best device was: Quadro P4000
INFO [11340] The following device was selected as main device: Quadro P4000
INFO [11340] Writing directly to 3D textures/images is NOT supported on main device
INFO [11340] Resizing window to 1024 1024
INFO [11340] Window not minimized; turning ON synchronized rendering
INFO [11340] Trying to start computation thread
INFO [11340] Computation thread started
D:/fastpath_temp/project.txt
D:/fastpath_temp/
INFO [11340] EXECUTING WholeSlideImageImporter because PO is modified.
INFO [11340] Metadata: openslide.comment =
10240x12288 (256x256) JPEG/RGB Q=80|Date = 26/11/2021|Time = 15:56:12|MPP = 0.24984|3dh_PixelSizeX = 0.249711|3dh_PixelSizeY = 0.249968|3dh_Filter = Default|3dh_Profile = Routine x20|3dh_ScannerHardwareID = SC150-210905
INFO [11340] Metadata: openslide.level-count = 8
INFO [11340] Metadata: openslide.level[0].downsample = 1
INFO [11340] Metadata: openslide.level[0].height = 12288
INFO [11340] Metadata: openslide.level[0].tile-height = 128
INFO [11340] Metadata: openslide.level[0].tile-width = 128
INFO [11340] Metadata: openslide.level[0].width = 10240
INFO [11340] Metadata: openslide.level[1].downsample = 2
INFO [11340] Metadata: openslide.level[1].height = 6144
INFO [11340] Metadata: openslide.level[1].tile-height = 128
INFO [11340] Metadata: openslide.level[1].tile-width = 128
INFO [11340] Metadata: openslide.level[1].width = 5120
INFO [11340] Metadata: openslide.level[2].downsample = 4
INFO [11340] Metadata: openslide.level[2].height = 3072
INFO [11340] Metadata: openslide.level[2].tile-height = 128
INFO [11340] Metadata: openslide.level[2].tile-width = 128
INFO [11340] Metadata: openslide.level[2].width = 2560
INFO [11340] Metadata: openslide.level[3].downsample = 8
INFO [11340] Metadata: openslide.level[3].height = 1536
INFO [11340] Metadata: openslide.level[3].tile-height = 128
INFO [11340] Metadata: openslide.level[3].tile-width = 128
INFO [11340] Metadata: openslide.level[3].width = 1280
INFO [11340] Metadata: openslide.level[4].downsample = 16
INFO [11340] Metadata: openslide.level[4].height = 768
INFO [11340] Metadata: openslide.level[4].tile-height = 128
INFO [11340] Metadata: openslide.level[4].tile-width = 128
INFO [11340] Metadata: openslide.level[4].width = 640
INFO [11340] Metadata: openslide.level[5].downsample = 32
INFO [11340] Metadata: openslide.level[5].height = 384
INFO [11340] Metadata: openslide.level[5].tile-height = 128
INFO [11340] Metadata: openslide.level[5].tile-width = 128
INFO [11340] Metadata: openslide.level[5].width = 320
INFO [11340] Metadata: openslide.level[6].downsample = 64
INFO [11340] Metadata: openslide.level[6].height = 192
INFO [11340] Metadata: openslide.level[6].tile-height = 128
INFO [11340] Metadata: openslide.level[6].tile-width = 128
INFO [11340] Metadata: openslide.level[6].width = 160
INFO [11340] Metadata: openslide.level[7].downsample = 128
INFO [11340] Metadata: openslide.level[7].height = 96
INFO [11340] Metadata: openslide.level[7].tile-height = 128
INFO [11340] Metadata: openslide.level[7].tile-width = 128
INFO [11340] Metadata: openslide.level[7].width = 80
INFO [11340] Metadata: openslide.quickhash-1 = f65e4e1293996bbe2c75dbb9b5012169822d702f8fdd9210540bd265237bb942
INFO [11340] Metadata: openslide.vendor = generic-tiff
INFO [11340] Metadata: tiff.ImageDescription =
10240x12288 (256x256) JPEG/RGB Q=80|Date = 26/11/2021|Time = 15:56:12|MPP = 0.24984|3dh_PixelSizeX = 0.249711|3dh_PixelSizeY = 0.249968|3dh_Filter = Default|3dh_Profile = Routine x20|3dh_ScannerHardwareID = SC150-210905
INFO [11340] Metadata: tiff.ResolutionUnit = centimeter
INFO [11340] Metadata: tiff.XResolution = 40046.31640625
INFO [11340] Metadata: tiff.YResolution = 40005.28515625
INFO [11340] WSI has 8 levels
INFO [11340] Checking level 0
INFO [11340] WSI level 0 has size 10240x12288 and 480 MB adding..
INFO [11340] Setting tile width and height to 128 128
INFO [11340] Checking level 1
INFO [11340] WSI level 1 has size 5120x6144 and 120 MB adding..
INFO [11340] Setting tile width and height to 128 128
INFO [11340] Checking level 2
INFO [11340] WSI level 2 has size 2560x3072 and 30 MB adding..
INFO [11340] Setting tile width and height to 128 128
INFO [11340] Checking level 3
INFO [11340] WSI level 3 has size 1280x1536 and 7 MB adding..
INFO [11340] Setting tile width and height to 128 128
INFO [11340] Checking level 4
INFO [11340] WSI level was less than 4 MB, skipping..
calling insert renderer
INFO [11340] EXECUTING EmptyProcessObject because PO is modified.
INFO [11340] EXECUTING ImagePyramidRenderer because PO is modified.
INFO [11340] Recalculating the camera of the view
INFO [11340] Finished initializing OpenGL
finished insert renderer
INFO [11340] Recalculating the camera of the view
INFO [11340] Finished initializing OpenGL
Current WSI used: D:/Pradeep/Andre/pyramidal_conv/pyramidal2.tif
Current result folder path: D:/fastpath_temp//results/pyramidal2
current file: D:/fastpath_temp//results/pyramidal2/.
current file: D:/fastpath_temp//results/pyramidal2/..
Path: D:/fastpath_temp/thumbnails/pyramidal2.png
Thumbnail does not exist! Creating one from the WSI
INFO [11340] EXECUTING EmptyProcessObject because PO is modified.
INFO [11340] EXECUTING ImageChannelConverter because PO is modified.
Model name in wrapper: Epithelium_HE_UNET_512
Final model metadata config sent to pixelClassifier:
m[IE] = (TensorRT)
m[batch_process] = (2)
m[batch_size] = (1)
m[class_colors] = (0,0,255;0,255,0)
m[class_names] = (Exterior;Epithelium)
m[cpu] = (0)
m[input_img_size_x] = (512)
m[input_img_size_y] = (512)
m[input_node] = (ImageInputLayer)
m[interpolation] = (0)
m[magnification_level] = (10)
m[mask_threshold] = (0.02)
m[model_name] = (Epithelium_HE_512)
m[name] = (Epithelium_HE_512)
m[nb_channels] = (3)
m[nb_classes] = (2)
m[output_node] = (Softmax_Layer_Transpose2)
m[patch_overlap] = (0.05)
m[pipeline] = (import;tissue_segmentation;batchgen;neural_network;stitch;render)
m[problem] = (segmentation)
m[resolution] = (high)
m[scale_factor] = (1.0f/1.0f)
m[task] = (Epithelium_HE_512)
m[tissue_threshold] = (70)
Final model metadata config WITHIN to pixelClassifier:
m[IE] = (TensorRT)
m[batch_process] = (2)
m[batch_size] = (1)
m[class_colors] = (0,0,255;0,255,0)
m[class_names] = (Exterior;Epithelium)
m[cpu] = (0)
m[input_img_size_x] = (512)
m[input_img_size_y] = (512)
m[input_node] = (ImageInputLayer)
m[interpolation] = (0)
m[magnification_level] = (10)
m[mask_threshold] = (0.02)
m[model_name] = (Epithelium_HE_512)
m[name] = (Epithelium_HE_512)
m[nb_channels] = (3)
m[nb_classes] = (2)
m[output_node] = (Softmax_Layer_Transpose2)
m[patch_overlap] = (0.05)
m[pipeline] = (import;tissue_segmentation;batchgen;neural_network;stitch;render)
m[problem] = (segmentation)
m[resolution] = (high)
m[scale_factor] = (1.0f/1.0f)
m[task] = (Epithelium_HE_512)
m[tissue_threshold] = (70)
Current model: Epithelium_HE_UNET_512
current WSI: D:/Pradeep/Andre/pyramidal_conv/pyramidal2.tif
Curr patch level: 2
current WSI: D:/Pradeep/Andre/pyramidal_conv/pyramidal2.tif
INFO [11340] EXECUTING WholeSlideImageImporter because PO is modified.
INFO [11340] Metadata: openslide.comment =
10240x12288 (256x256) JPEG/RGB Q=80|Date = 26/11/2021|Time = 15:56:12|MPP = 0.24984|3dh_PixelSizeX = 0.249711|3dh_PixelSizeY = 0.249968|3dh_Filter = Default|3dh_Profile = Routine x20|3dh_ScannerHardwareID = SC150-210905
INFO [11340] Metadata: openslide.level-count = 8
INFO [11340] Metadata: openslide.level[0].downsample = 1
INFO [11340] Metadata: openslide.level[0].height = 12288
INFO [11340] Metadata: openslide.level[0].tile-height = 128
INFO [11340] Metadata: openslide.level[0].tile-width = 128
INFO [11340] Metadata: openslide.level[0].width = 10240
INFO [11340] Metadata: openslide.level[1].downsample = 2
INFO [11340] Metadata: openslide.level[1].height = 6144
INFO [11340] Metadata: openslide.level[1].tile-height = 128
INFO [11340] Metadata: openslide.level[1].tile-width = 128
INFO [11340] Metadata: openslide.level[1].width = 5120
INFO [11340] Metadata: openslide.level[2].downsample = 4
INFO [11340] Metadata: openslide.level[2].height = 3072
INFO [11340] Metadata: openslide.level[2].tile-height = 128
INFO [11340] Metadata: openslide.level[2].tile-width = 128
INFO [11340] Metadata: openslide.level[2].width = 2560
INFO [11340] Metadata: openslide.level[3].downsample = 8
INFO [11340] Metadata: openslide.level[3].height = 1536
INFO [11340] Metadata: openslide.level[3].tile-height = 128
INFO [11340] Metadata: openslide.level[3].tile-width = 128
INFO [11340] Metadata: openslide.level[3].width = 1280
INFO [11340] Metadata: openslide.level[4].downsample = 16
INFO [11340] Metadata: openslide.level[4].height = 768
INFO [11340] Metadata: openslide.level[4].tile-height = 128
INFO [11340] Metadata: openslide.level[4].tile-width = 128
INFO [11340] Metadata: openslide.level[4].width = 640
INFO [11340] Metadata: openslide.level[5].downsample = 32
INFO [11340] Metadata: openslide.level[5].height = 384
INFO [11340] Metadata: openslide.level[5].tile-height = 128
INFO [11340] Metadata: openslide.level[5].tile-width = 128
INFO [11340] Metadata: openslide.level[5].width = 320
INFO [11340] Metadata: openslide.level[6].downsample = 64
INFO [11340] Metadata: openslide.level[6].height = 192
INFO [11340] Metadata: openslide.level[6].tile-height = 128
INFO [11340] Metadata: openslide.level[6].tile-width = 128
INFO [11340] Metadata: openslide.level[6].width = 160
INFO [11340] Metadata: openslide.level[7].downsample = 128
INFO [11340] Metadata: openslide.level[7].height = 96
INFO [11340] Metadata: openslide.level[7].tile-height = 128
INFO [11340] Metadata: openslide.level[7].tile-width = 128
INFO [11340] Metadata: openslide.level[7].width = 80
INFO [11340] Metadata: openslide.quickhash-1 = f65e4e1293996bbe2c75dbb9b5012169822d702f8fdd9210540bd265237bb942
INFO [11340] Metadata: openslide.vendor = generic-tiff
INFO [11340] Metadata: tiff.ImageDescription =
10240x12288 (256x256) JPEG/RGB Q=80|Date = 26/11/2021|Time = 15:56:12|MPP = 0.24984|3dh_PixelSizeX = 0.249711|3dh_PixelSizeY = 0.249968|3dh_Filter = Default|3dh_Profile = Routine x20|3dh_ScannerHardwareID = SC150-210905
INFO [11340] Metadata: tiff.ResolutionUnit = centimeter
INFO [11340] Metadata: tiff.XResolution = 40046.31640625
INFO [11340] Metadata: tiff.YResolution = 40005.28515625
INFO [11340] WSI has 8 levels
INFO [11340] Checking level 0
INFO [11340] WSI level 0 has size 10240x12288 and 480 MB adding..
INFO [11340] Setting tile width and height to 128 128
INFO [11340] Checking level 1
INFO [11340] WSI level 1 has size 5120x6144 and 120 MB adding..
INFO [11340] Setting tile width and height to 128 128
INFO [11340] Checking level 2
INFO [11340] WSI level 2 has size 2560x3072 and 30 MB adding..
INFO [11340] Setting tile width and height to 128 128
INFO [11340] Checking level 3
INFO [11340] WSI level 3 has size 1280x1536 and 7 MB adding..
INFO [11340] Setting tile width and height to 128 128
INFO [11340] Checking level 4
INFO [11340] WSI level was less than 4 MB, skipping..
Current OS is: windows
Current kernel is: winnt
accepted models: .onnx
accepted models: .txt
INFO [11340] Loading inference engines in folder C:/Program Files/FastPathology/bin//..//bin/
INFO [11340] Loading inference engine OpenVINO from shared library InferenceEngineOpenVINO.dll
INFO [11340] Loading inference engine TensorFlow from shared library InferenceEngineTensorFlow.dll
2021-12-10 19:14:24.157437: I tensorflow/stream_executor/platform/default/dso_loader.cc:49] Successfully opened dynamic library cudart64_110.dll
INFO [11340] Loading inference engine TensorRT from shared library InferenceEngineTensorRT.dll
INFO [11340] Inference engine TensorRT selected
INFO [11340] Inference engine TensorRT selected
Current available IEs:
OpenVINO,
TensorFlow,
TensorRT,
Which model formats are available and that there exists an IE for:
.onnx,
.txt,
TensorRT (using ONNX) selected
INFO [11340] Inference engine TensorRT selected
Model was found.
Preselected IE was used: TensorRT
INFO [11340] Inference engine TensorRT selected
INFO [11340] Serialized file C:/ProgramData/FAST/kernel_binaries///Epithelium_HE_UNET_512.onnx_653401940713773443.bin is up to date.
INFO [11340] [MemUsageChange] Init CUDA: CPU +267, GPU +0, now: CPU 13562, GPU 1146 (MiB)
INFO [11340] Loaded engine size: 1245 MiB
WARNING [11340] TensorRT was linked against cuBLAS/cuBLAS LT 11.6.3 but loaded cuBLAS/cuBLAS LT 11.4.1
INFO [11340] [MemUsageChange] Init cuBLAS/cuBLASLt: CPU +224, GPU +76, now: CPU 15043, GPU 2468 (MiB)
INFO [11340] [MemUsageChange] Init cuDNN: CPU +252, GPU +78, now: CPU 15294, GPU 2546 (MiB)
INFO [11340] [MemUsageChange] TensorRT-managed allocation in engine deserialization: CPU +0, GPU +1245, now: CPU 0, GPU 1245 (MiB)
INFO [11340] Guessed image ordering to be channel first as shape was 1 3 512 512
INFO [11340] Found input node ImageInputLayer with shape 1 3 512 512
INFO [11340] Guessed image ordering to be channel first as shape was 1 2 512 512
INFO [11340] Found output node Softmax_Layer_Transpose2 with shape 1 2 512 512
WARNING [11340] TensorRT was linked against cuBLAS/cuBLAS LT 11.6.3 but loaded cuBLAS/cuBLAS LT 11.4.1
INFO [11340] [MemUsageChange] Init cuBLAS/cuBLASLt: CPU +3, GPU +10, now: CPU 14047, GPU 2538 (MiB)
INFO [11340] [MemUsageChange] Init cuDNN: CPU +2, GPU +8, now: CPU 14049, GPU 2546 (MiB)
INFO [11340] [MemUsageChange] TensorRT-managed allocation in IExecutionContext creation: CPU +0, GPU +1362, now: CPU 0, GPU 2607 (MiB)
Threshold was defined: 70
tissue_threshold was defined, so is performing thresholding as preprocessing step.
Setting mask_threshold to: 0.02
calling insert renderer
INFO [11340] EXECUTING EmptyProcessObject because PO is modified.
INFO [11340] EXECUTING TissueSegmentation because PO is modified.
INFO [11340] EXECUTING EmptyProcessObject because PO is modified.
INFO [11340] EXECUTING ImageChannelConverter because PO is modified.
INFO [11340] EXECUTING EmptyProcessObject because PO is modified.
INFO [11340] EXECUTING Dilation because PO is modified.
INFO [11340] EXECUTING Erosion because PO is modified.
INFO [11340] EXECUTING EmptyProcessObject because PO is modified.
INFO [11340] EXECUTING PatchGenerator because PO is modified.
WARNING [22388] Patch size must be a multiple of 16 (TIFF limitation). Adding some overlap (32, 32) to fix.
INFO [22388] Generating patch 2 1
INFO [22388] EXECUTING EmptyProcessObject because PO is modified.
INFO [22388] EXECUTING ImageChannelConverter because PO is modified.
INFO [22388] Generating patch 3 1
INFO [22388] EXECUTING EmptyProcessObject because PO is modified.
INFO [22388] EXECUTING ImageChannelConverter because PO is modified.
INFO [11340] EXECUTING SegmentationNetwork because PO is modified.
INFO [22388] Generating patch 1 2
INFO [11340] Finished copying input data to TensorRT
INFO [22388] EXECUTING EmptyProcessObject because PO is modified.
INFO [22388] EXECUTING ImageChannelConverter because PO is modified.
INFO [22388] Generating patch 2 2
INFO [11340] Finished execute TensorRT
INFO [11340] Finished freeing input data TensorRTINFO [22388] EXECUTING EmptyProcessObject because PO is modified.
INFO [11340] Processing output node Softmax_Layer_Transpose2

INFO [22388] EXECUTING ImageChannelConverter because PO is modified.INFO [11340] Finished moving data to FAST tensor, TensorRT

INFO [11340] Finished transfer of output data TensorRT
INFO [11340] Finished freeing output data TensorRTINFO [22388] Generating patch 3 2

INFO [11340] EXECUTING EmptyProcessObject because PO is modified.
INFO [11340] EXECUTING TensorToSegmentation because PO is modified.
INFO [11340] EXECUTING PatchStitcher because PO is modified.
INFO [11340] Patch stitcher creating image with size 2560 3072
INFO [11340] Stitching 2D data 2 1
INFO [11340] EXECUTING SegmentationRenderer because PO is modified.
INFO [11340] Recalculating the camera of the view
INFO [11340] Finished initializing OpenGLINFO [22388] EXECUTING EmptyProcessObject because PO is modified.
finished insert renderer
Current save location:
D:/fastpath_temp/
INFO [22388] EXECUTING ImageChannelConverter because PO is modified.
INFO [22388] Generating patch 4 2
INFO [14232] EXECUTING SegmentationNetwork because PO has new input data.
INFO [22388] EXECUTING EmptyProcessObject because PO is modified.
INFO [22388] EXECUTING ImageChannelConverter because PO is modified.
INFO [22388] Generating patch 5 2
INFO [14232] Finished copying input data to TensorRT
INFO [22388] EXECUTING EmptyProcessObject because PO is modified.
INFO [22388] EXECUTING ImageChannelConverter because PO is modified.
INFO [22388] Generating patch 1 3
Inference thread is finished...
INFO [14232] Finished execute TensorRT
INFO [14232] Finished freeing input data TensorRT
INFO [14232] Processing output node Softmax_Layer_Transpose2
INFO [14232] Finished moving data to FAST tensor, TensorRT
INFO [14232] Finished transfer of output data TensorRT
INFO [14232] Finished freeing output data TensorRT
INFO [14232] EXECUTING EmptyProcessObject because PO is modified.
INFO [14232] EXECUTING TensorToSegmentation because PO is modified.
INFO [14232] EXECUTING PatchStitcher because PO has new input data.
INFO [14232] Stitching 2D data 3 1
INFO [14232] EXECUTING SegmentationNetwork because PO has new input data.
INFO [14232] Finished copying input data to TensorRT
INFO [22388] EXECUTING EmptyProcessObject because PO is modified.
INFO [22388] EXECUTING ImageChannelConverter because PO is modified.
INFO [22388] Generating patch 2 3
INFO [14232] Finished execute TensorRT
INFO [14232] Finished freeing input data TensorRT
INFO [14232] Processing output node Softmax_Layer_Transpose2
INFO [22388] EXECUTING EmptyProcessObject because PO is modified.INFO [14232] Finished moving data to FAST tensor, TensorRT
INFO [22388] EXECUTING ImageChannelConverter because PO is modified.
INFO [14232] Finished transfer of output data TensorRT

INFO [14232] Finished freeing output data TensorRTINFO [22388] Generating patch 3 3

INFO [14232] EXECUTING EmptyProcessObject because PO is modified.
INFO [14232] EXECUTING TensorToSegmentation because PO is modified.
INFO [14232] EXECUTING PatchStitcher because PO has new input data.
INFO [14232] Stitching 2D data 1 2
INFO [22388] EXECUTING EmptyProcessObject because PO is modified.
INFO [22388] EXECUTING ImageChannelConverter because PO is modified.INFO [14232] EXECUTING SegmentationNetwork because PO has new input data.

INFO [22388] Generating patch 4 3
INFO [14232] Finished copying input data to TensorRT
INFO [22388] EXECUTING EmptyProcessObject because PO is modified.
INFO [22388] EXECUTING ImageChannelConverter because PO is modified.
INFO [22388] Generating patch 5 3
INFO [22388] EXECUTING EmptyProcessObject because PO is modified.
INFO [22388] EXECUTING ImageChannelConverter because PO is modified.
INFO [14232] Finished execute TensorRTINFO [22388] Generating patch 1 4

INFO [14232] Finished freeing input data TensorRT
INFO [14232] Processing output node Softmax_Layer_Transpose2
INFO [14232] Finished moving data to FAST tensor, TensorRT
INFO [14232] Finished transfer of output data TensorRT
INFO [14232] Finished freeing output data TensorRT
INFO [14232] EXECUTING EmptyProcessObject because PO is modified.
INFO [14232] EXECUTING TensorToSegmentation because PO is modified.
INFO [14232] EXECUTING PatchStitcher because PO has new input data.
INFO [14232] Stitching 2D data 2 2INFO [22388] EXECUTING EmptyProcessObject because PO is modified.

INFO [22388] EXECUTING ImageChannelConverter because PO is modified.INFO [14232] EXECUTING SegmentationNetwork because PO has new input data.

INFO [22388] Generating patch 2 4
INFO [14232] Finished copying input data to TensorRT
INFO [22388] EXECUTING EmptyProcessObject because PO is modified.
INFO [22388] EXECUTING ImageChannelConverter because PO is modified.
INFO [22388] Generating patch 3 4
INFO [14232] Finished execute TensorRT
INFO [14232] Finished freeing input data TensorRT
INFO [14232] Processing output node Softmax_Layer_Transpose2
INFO [14232] Finished moving data to FAST tensor, TensorRT
INFO [14232] Finished transfer of output data TensorRT
INFO [14232] Finished freeing output data TensorRT
INFO [22388] EXECUTING EmptyProcessObject because PO is modified.INFO [14232] EXECUTING EmptyProcessObject because PO is modified.
INFO [22388] EXECUTING ImageChannelConverter because PO is modified.

INFO [14232] EXECUTING TensorToSegmentation because PO is modified.INFO [22388] Generating patch 4 4

INFO [14232] EXECUTING PatchStitcher because PO has new input data.
INFO [14232] Stitching 2D data 3 2
INFO [14232] EXECUTING SegmentationNetwork because PO has new input data.
INFO [14232] Finished copying input data to TensorRT
INFO [22388] EXECUTING EmptyProcessObject because PO is modified.
INFO [22388] EXECUTING ImageChannelConverter because PO is modified.
INFO [22388] Generating patch 5 4
INFO [22388] EXECUTING EmptyProcessObject because PO is modified.
INFO [22388] EXECUTING ImageChannelConverter because PO is modified.
INFO [14232] Finished execute TensorRTINFO [22388] Generating patch 1 5

INFO [14232] Finished freeing input data TensorRT
INFO [14232] Processing output node Softmax_Layer_Transpose2
INFO [14232] Finished moving data to FAST tensor, TensorRT
INFO [14232] Finished transfer of output data TensorRT
INFO [14232] Finished freeing output data TensorRT
INFO [14232] EXECUTING EmptyProcessObject because PO is modified.
INFO [14232] EXECUTING TensorToSegmentation because PO is modified.
INFO [14232] EXECUTING PatchStitcher because PO has new input data.
INFO [14232] Stitching 2D data 4 2
INFO [14232] EXECUTING SegmentationNetwork because PO has new input data.INFO [22388] EXECUTING EmptyProcessObject because PO is modified.

INFO [22388] EXECUTING ImageChannelConverter because PO is modified.INFO [14232] EXECUTING EmptyProcessObject because PO is modified.

INFO [14232] EXECUTING ImageResizer because PO is modified.
INFO [22388] Generating patch 2 5
INFO [14232] Finished copying input data to TensorRT
INFO [22388] EXECUTING EmptyProcessObject because PO is modified.
INFO [22388] EXECUTING ImageChannelConverter because PO is modified.
INFO [22388] Generating patch 3 5
INFO [14232] Finished execute TensorRT
INFO [14232] Finished freeing input data TensorRTINFO [22388] EXECUTING EmptyProcessObject because PO is modified.

INFO [14232] Processing output node Softmax_Layer_Transpose2INFO [22388] EXECUTING ImageChannelConverter because PO is modified.

INFO [14232] Finished moving data to FAST tensor, TensorRTINFO [
22388] Generating patch 4 5INFO [14232] Finished transfer of output data TensorRT

INFO [14232] Finished freeing output data TensorRT
INFO [14232] EXECUTING EmptyProcessObject because PO is modified.
INFO [14232] EXECUTING TensorToSegmentation because PO is modified.
INFO [14232] EXECUTING PatchStitcher because PO has new input data.
INFO [14232] Stitching 2D data 5 2
INFO [14232] EXECUTING SegmentationNetwork because PO has new input data.
INFO [14232] Finished copying input data to TensorRTINFO [22388] EXECUTING EmptyProcessObject because PO is modified.

INFO [22388] EXECUTING ImageChannelConverter because PO is modified.
INFO [22388] Generating patch 5 5
INFO [22388] EXECUTING EmptyProcessObject because PO is modified.
INFO [22388] EXECUTING ImageChannelConverter because PO is modified.
INFO [14232] Finished execute TensorRT
INFO [22388] Generating patch 2 6INFO [14232]
Finished freeing input data TensorRT
INFO [14232] Processing output node Softmax_Layer_Transpose2
INFO [14232] Finished moving data to FAST tensor, TensorRT
INFO [14232] Finished transfer of output data TensorRT
INFO [14232] Finished freeing output data TensorRT
INFO [14232] EXECUTING EmptyProcessObject because PO is modified.INFO [22388] EXECUTING EmptyProcessObject because PO is modified.

INFO [14232] EXECUTING TensorToSegmentation because PO is modified.INFO [22388] EXECUTING ImageChannelConverter because PO is modified.

INFO [14232] EXECUTING PatchStitcher because PO has new input data.INFO [22388] Generating patch 3 6

INFO [14232] Stitching 2D data 1 3
INFO [14232] EXECUTING SegmentationNetwork because PO has new input data.
INFO [14232] Finished copying input data to TensorRTINFO [22388] EXECUTING EmptyProcessObject because PO is modified.

INFO [22388] EXECUTING ImageChannelConverter because PO is modified.
INFO [22388] Done generating patches
INFO [14232] Finished execute TensorRT
INFO [14232] Finished freeing input data TensorRT
INFO [14232] Processing output node Softmax_Layer_Transpose2
INFO [14232] Finished moving data to FAST tensor, TensorRT
INFO [14232] Finished transfer of output data TensorRT
INFO [14232] Finished freeing output data TensorRT
INFO [14232] EXECUTING EmptyProcessObject because PO is modified.
INFO [14232] EXECUTING TensorToSegmentation because PO is modified.
INFO [14232] EXECUTING PatchStitcher because PO has new input data.
INFO [14232] Stitching 2D data 2 3
INFO [14232] EXECUTING SegmentationNetwork because PO has new input data.
INFO [14232] Finished copying input data to TensorRT
INFO [14232] Finished execute TensorRT
INFO [14232] Finished freeing input data TensorRT
INFO [14232] Processing output node Softmax_Layer_Transpose2
INFO [14232] Finished moving data to FAST tensor, TensorRT
INFO [14232] Finished transfer of output data TensorRT
INFO [14232] Finished freeing output data TensorRT
INFO [14232] EXECUTING EmptyProcessObject because PO is modified.
INFO [14232] EXECUTING TensorToSegmentation because PO is modified.
INFO [14232] EXECUTING PatchStitcher because PO has new input data.
INFO [14232] Stitching 2D data 3 3
INFO [14232] EXECUTING SegmentationNetwork because PO has new input data.
INFO [14232] Finished copying input data to TensorRT
INFO [14232] Finished execute TensorRT
INFO [14232] Finished freeing input data TensorRT
INFO [14232] Processing output node Softmax_Layer_Transpose2
INFO [14232] Finished moving data to FAST tensor, TensorRT
INFO [14232] Finished transfer of output data TensorRT
INFO [14232] Finished freeing output data TensorRT
INFO [14232] EXECUTING EmptyProcessObject because PO is modified.
INFO [14232] EXECUTING TensorToSegmentation because PO is modified.
INFO [14232] EXECUTING PatchStitcher because PO has new input data.
INFO [14232] Stitching 2D data 4 3
INFO [14232] EXECUTING SegmentationNetwork because PO has new input data.
INFO [14232] EXECUTING EmptyProcessObject because PO is modified.
INFO [14232] EXECUTING ImageResizer because PO is modified.
INFO [14232] Finished copying input data to TensorRT
INFO [14232] Finished execute TensorRT
INFO [14232] Finished freeing input data TensorRT
INFO [14232] Processing output node Softmax_Layer_Transpose2
INFO [14232] Finished moving data to FAST tensor, TensorRT
INFO [14232] Finished transfer of output data TensorRT
INFO [14232] Finished freeing output data TensorRT
INFO [14232] EXECUTING EmptyProcessObject because PO is modified.
INFO [14232] EXECUTING TensorToSegmentation because PO is modified.
INFO [14232] EXECUTING PatchStitcher because PO has new input data.
INFO [14232] Stitching 2D data 5 3
INFO [14232] EXECUTING SegmentationNetwork because PO has new input data.
INFO [14232] Finished copying input data to TensorRT
INFO [14232] Finished execute TensorRT
INFO [14232] Finished freeing input data TensorRT
INFO [14232] Processing output node Softmax_Layer_Transpose2
INFO [14232] Finished moving data to FAST tensor, TensorRT
INFO [14232] Finished transfer of output data TensorRT
INFO [14232] Finished freeing output data TensorRT
INFO [14232] EXECUTING EmptyProcessObject because PO is modified.
INFO [14232] EXECUTING TensorToSegmentation because PO is modified.
INFO [14232] EXECUTING PatchStitcher because PO has new input data.
INFO [14232] Stitching 2D data 1 4
INFO [14232] EXECUTING SegmentationNetwork because PO has new input data.
INFO [14232] Finished copying input data to TensorRT
INFO [14232] Finished execute TensorRT
INFO [14232] Finished freeing input data TensorRT
INFO [14232] Processing output node Softmax_Layer_Transpose2
INFO [14232] Finished moving data to FAST tensor, TensorRT
INFO [14232] Finished transfer of output data TensorRT
INFO [14232] Finished freeing output data TensorRT
INFO [14232] EXECUTING EmptyProcessObject because PO is modified.
INFO [14232] EXECUTING TensorToSegmentation because PO is modified.
INFO [14232] EXECUTING PatchStitcher because PO has new input data.
INFO [14232] Stitching 2D data 2 4
INFO [14232] EXECUTING SegmentationNetwork because PO has new input data.
INFO [14232] Finished copying input data to TensorRT
INFO [14232] Finished execute TensorRT
INFO [14232] Finished freeing input data TensorRT
INFO [14232] Processing output node Softmax_Layer_Transpose2
INFO [14232] Finished moving data to FAST tensor, TensorRT
INFO [14232] Finished transfer of output data TensorRT
INFO [14232] Finished freeing output data TensorRT
INFO [14232] EXECUTING EmptyProcessObject because PO is modified.
INFO [14232] EXECUTING TensorToSegmentation because PO is modified.
INFO [14232] EXECUTING PatchStitcher because PO has new input data.
INFO [14232] Stitching 2D data 3 4
INFO [14232] EXECUTING SegmentationNetwork because PO has new input data.
INFO [14232] Finished copying input data to TensorRT
INFO [14232] Finished execute TensorRT
INFO [14232] Finished freeing input data TensorRT
INFO [14232] Processing output node Softmax_Layer_Transpose2
INFO [14232] Finished moving data to FAST tensor, TensorRT
INFO [14232] Finished transfer of output data TensorRT
INFO [14232] Finished freeing output data TensorRT
INFO [14232] EXECUTING EmptyProcessObject because PO is modified.
INFO [14232] EXECUTING TensorToSegmentation because PO is modified.
INFO [14232] EXECUTING PatchStitcher because PO has new input data.
INFO [14232] Stitching 2D data 4 4
INFO [14232] EXECUTING SegmentationNetwork because PO has new input data.
INFO [14232] EXECUTING EmptyProcessObject because PO is modified.
INFO [14232] EXECUTING ImageResizer because PO is modified.
INFO [14232] Finished copying input data to TensorRT
INFO [14232] Finished execute TensorRT
INFO [14232] Finished freeing input data TensorRT
INFO [14232] Processing output node Softmax_Layer_Transpose2
INFO [14232] Finished moving data to FAST tensor, TensorRT
INFO [14232] Finished transfer of output data TensorRT
INFO [14232] Finished freeing output data TensorRT
INFO [14232] EXECUTING EmptyProcessObject because PO is modified.
INFO [14232] EXECUTING TensorToSegmentation because PO is modified.
INFO [14232] EXECUTING PatchStitcher because PO has new input data.
INFO [14232] Stitching 2D data 5 4
INFO [14232] EXECUTING SegmentationNetwork because PO has new input data.
INFO [14232] Finished copying input data to TensorRT
INFO [14232] Finished execute TensorRT
INFO [14232] Finished freeing input data TensorRT
INFO [14232] Processing output node Softmax_Layer_Transpose2
INFO [14232] Finished moving data to FAST tensor, TensorRT
INFO [14232] Finished transfer of output data TensorRT
INFO [14232] Finished freeing output data TensorRT
INFO [14232] EXECUTING EmptyProcessObject because PO is modified.
INFO [14232] EXECUTING TensorToSegmentation because PO is modified.
INFO [14232] EXECUTING PatchStitcher because PO has new input data.
INFO [14232] Stitching 2D data 1 5
INFO [14232] EXECUTING SegmentationNetwork because PO has new input data.
INFO [14232] Finished copying input data to TensorRT
INFO [14232] Finished execute TensorRT
INFO [14232] Finished freeing input data TensorRT
INFO [14232] Processing output node Softmax_Layer_Transpose2
INFO [14232] Finished moving data to FAST tensor, TensorRT
INFO [14232] Finished transfer of output data TensorRT
INFO [14232] Finished freeing output data TensorRT
INFO [14232] EXECUTING EmptyProcessObject because PO is modified.
INFO [14232] EXECUTING TensorToSegmentation because PO is modified.
INFO [14232] EXECUTING PatchStitcher because PO has new input data.
INFO [14232] Stitching 2D data 2 5
INFO [14232] EXECUTING SegmentationNetwork because PO has new input data.
INFO [14232] Finished copying input data to TensorRT
INFO [14232] Finished execute TensorRT
INFO [14232] Finished freeing input data TensorRT
INFO [14232] Processing output node Softmax_Layer_Transpose2
INFO [14232] Finished moving data to FAST tensor, TensorRT
INFO [14232] Finished transfer of output data TensorRT
INFO [14232] Finished freeing output data TensorRT
INFO [14232] EXECUTING EmptyProcessObject because PO is modified.
INFO [14232] EXECUTING TensorToSegmentation because PO is modified.
INFO [14232] EXECUTING PatchStitcher because PO has new input data.
INFO [14232] Stitching 2D data 3 5
INFO [14232] EXECUTING SegmentationNetwork because PO has new input data.
INFO [14232] Finished copying input data to TensorRT
INFO [14232] Finished execute TensorRT
INFO [14232] Finished freeing input data TensorRT
INFO [14232] Processing output node Softmax_Layer_Transpose2
INFO [14232] Finished moving data to FAST tensor, TensorRT
INFO [14232] Finished transfer of output data TensorRT
INFO [14232] Finished freeing output data TensorRT
INFO [14232] EXECUTING EmptyProcessObject because PO is modified.
INFO [14232] EXECUTING TensorToSegmentation because PO is modified.
INFO [14232] EXECUTING PatchStitcher because PO has new input data.
INFO [14232] Stitching 2D data 4 5
INFO [14232] EXECUTING SegmentationNetwork because PO has new input data.
INFO [14232] EXECUTING EmptyProcessObject because PO is modified.
INFO [14232] EXECUTING ImageResizer because PO is modified.
INFO [14232] Finished copying input data to TensorRT
INFO [14232] Finished execute TensorRT
INFO [14232] Finished freeing input data TensorRT
INFO [14232] Processing output node Softmax_Layer_Transpose2
INFO [14232] Finished moving data to FAST tensor, TensorRT
INFO [14232] Finished transfer of output data TensorRT
INFO [14232] Finished freeing output data TensorRT
INFO [14232] EXECUTING EmptyProcessObject because PO is modified.
INFO [14232] EXECUTING TensorToSegmentation because PO is modified.
INFO [14232] EXECUTING PatchStitcher because PO has new input data.
INFO [14232] Stitching 2D data 5 5
INFO [14232] EXECUTING SegmentationNetwork because PO has new input data.
INFO [14232] EXECUTING EmptyProcessObject because PO is modified.
INFO [14232] EXECUTING ImageResizer because PO is modified.
INFO [14232] Finished copying input data to TensorRT
INFO [14232] Finished execute TensorRT
INFO [14232] Finished freeing input data TensorRT
INFO [14232] Processing output node Softmax_Layer_Transpose2
INFO [14232] Finished moving data to FAST tensor, TensorRT
INFO [14232] Finished transfer of output data TensorRT
INFO [14232] Finished freeing output data TensorRT
INFO [14232] EXECUTING EmptyProcessObject because PO is modified.
INFO [14232] EXECUTING TensorToSegmentation because PO is modified.
INFO [14232] EXECUTING PatchStitcher because PO has new input data.
INFO [14232] Stitching 2D data 2 6
INFO [14232] EXECUTING SegmentationNetwork because PO has new input data.
INFO [14232] EXECUTING EmptyProcessObject because PO is modified.
INFO [14232] EXECUTING ImageResizer because PO is modified.
INFO [14232] Finished copying input data to TensorRT
INFO [14232] Finished execute TensorRT
INFO [14232] Finished freeing input data TensorRT
INFO [14232] Processing output node Softmax_Layer_Transpose2
INFO [14232] Finished moving data to FAST tensor, TensorRT
INFO [14232] Finished transfer of output data TensorRT
INFO [14232] Finished freeing output data TensorRT
INFO [14232] EXECUTING EmptyProcessObject because PO is modified.
INFO [14232] EXECUTING TensorToSegmentation because PO is modified.
INFO [14232] EXECUTING PatchStitcher because PO has new input data.
INFO [14232] Stitching 2D data 3 6
INFO [14232] EXECUTING TIFFImagePyramidExporter because PO is modified.
INFO [14232] Data given to TIFFImagePyramidExporter was an Image, not an ImagePyramid, converting ...

[andreped]

I ran into the same issue again.
I created a project and ran the inference, which crashed..
I reopened the project and ran it again. Still had the same issue..

Hmm, interesting. I have observed that in the current build of FastPathology, but I believe this was working just fine on the version that was released.

Let me try to run some tests using the released version(s), and I will report what I observe.

NOTE: I accidentally clicked the wrong button. The issue will remain open until this problem has been solved.

[andreped]

BTW: Were you using a local WSI for this test? I saw that you used the WSI with path "D:/Pradeep/Andre/pyramidal_conv/pyramidal2.tif".

Could you try downloading and testing inference on a WSI from the IBDColEpi dataset? Try downloading the file "WSI_part_02.zip" (it is the smallest zip). Would be nice to know if there is something specific with the WSI that produces this issue when exporting (for whatever reason).

[pr4deepr]

Thanks a lot. I'll give this a go and report back.
Yes I've been using Windows 10..

[andreped]

Oh! Interesting observation. I downloaded and imported the 512 HE ONNX model from GitHub, and tested it on WSIs of very different sizes. It works well for most WSIs, especially larger WSIs. However, when testing the model on a particularly small WSI, it crashes here as well! Good catch, @pr4deepr !

That means there is something unstable during the exporter, that is when this is happening:

Data given to TIFFImagePyramidExporter was an Image, not an ImagePyramid, converting ...

I was not aware of this. We have not previously observed this issue. I tested on the WSI ID-30_HE_inactive.ndpi. I uploaded it to the Google Drive as well:
https://drive.google.com/drive/u/0/folders/1F591dj-d2xAXIMsKqm_qREHoaLEUDPW4

Please, test the pipeline on that WSI just to see that the full pipeline works. Then we will have to look into why the exporter fails on some WSIs (very strange...)

[pr4deepr]

Hi @smistad

You can find a dataset here.
https://cloudstor.aarnet.edu.au/plus/s/bZr6zQ9qI9GB1uZ

I have downscaled the images and they are originally from: ID-114_HE_inactive.ndpi in the google drive shared above.
large_pyramidal.tiff works on fastpathology
small_pyramidal.tiff crashes
This is when I use the Epithelial H&E model.

Cheers
Pradeep

[pr4deepr]

Actually, another problem. After I run the prediction the mask looks offset and edges of images are not predicted.

[andreped]

I have observed the same. I believe it might be a spacing issue with the importer, as it works well when importing the predictions in QuPath, using the old import script, right? Which WSI did you use? You will likely see the same for the IBDColEpi dataset. If you have a WSI @smistad could use for testing, he can try to find a fix for that. Regarding predictions on the left hand side? Did that happen during inference or during import? If during inference, try zooming a little on the region of interest. Might be that it failed to render it. It might also be an issue with the parameters you chose for the model. If you send me the WSI, I could see if I observe the same, and if there is an easy fix for it. If during import only, and it renders well in QuPath, then it is related to import/reader.

…

On Tue, 14 Dec 2021 at 03:03, Pradeep Rajasekhar ***@***.***> wrote: Actually, another problem. After I run the prediction the mask looks offset and edges of images are not predicted. [image: image] <https://user-images.githubusercontent.com/13831458/145919415-abd78dd7-5453-4c74-880b-2ce72f410e58.png> — You are receiving this because you modified the open/close state. Reply to this email directly, view it on GitHub <#8 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AG56G2JKFFKHGQOU33CYGRTUQ2QWTANCNFSM5JVUBVIA> .

[pr4deepr]

The predictions on the left are from inference in FastPathology .
It was the same even after importing into QuPath ..
I used a WSI from our facility. I'll try to get one tomorrow.

[andreped]

@pr4deepr Oh, that is interesting... Then there is something strange going on.

Have you tried just opening a single image and running inference on it? Do you get different results when adding multiple images to a project? Then it might be project-related.

You could test the test data to see if you observe the same there:
https://drive.google.com/drive/folders/1F591dj-d2xAXIMsKqm_qREHoaLEUDPW4?usp=sharing

[andreped]

@smistad Remember if you add padding, we should be able to specify which value to pad with, as for ultrasound you probably want to pad with 0, but for histopathological images you likely want to pad with 255.

Also, will this affect the low-resolution pipeline? Where we extract a low-res WSI and send that directly and send that to the network. Here it is useful that the input image is resized, if it does not match the expected NN input size. See here for what I mean. Perhaps, you could add this as an option? Whether to pad or resize? Might be most optimal.

[andreped]

@smistad I will add these fixes to the upcoming release of FP.