krematas / soccerontable Goto Github PK

I always failed in manual calibration, though I have set 4 correspondences.

The dataset is "barcelona" provided in your project.

Hello,thanks for sharing this project.
I got error when i want to excute simple_visualization.
The error stack like:

X Error of failed request: BadRequest (invalid request code or no su operation)
    Major opcode of failed request:１４９（RANDR）
    Minor opcode of failed request:8(RRGetScreenResources)
    Serial number of failed request:14
    Current serial number in output stream:14

I check the code ,it thrown out when python invoke the app.window() of glumpy moudle.
It seems like some incorrect configuration about driver. I have no idea.
Have you ever face this problem before? or do you have any idea?
Best wishes.

Hi, Thank you for your nice work? When I diving into the code, I found you there's 0.5 in unproject function as the depth. Could you please explain why you choose 0.5 as the depth value? Thank you!

Hi, thanks for sharing this great project.

I'm having this error when I run the estimate_poses.py

Traceback (most recent call last):
File "estimate_poses.py", line 19, in
db.estimate_poses(openpose_dir=opt.openpose_dir)
File "/home/lrvc/DeepStereo/soccer/soccerontable/soccer3d/core.py", line 270, in estimate_poses
with open(join(join(self.path_to_dataset, 'tmp'), '{0}_pose.yml'.format(j))) as data_file:
IOError: [Errno 2] No such file or directory: '/home/lrvc/DeepStereo/soccer/dataset/barcelona/tmp/0_pose.yml'

and I couldn't find anywhere in the code where it creates the file 0_pose.yml. Is this in somewhat related to the openpose installation?

thanks for the help
best

Using command python3 demo/calibrate_video.py --path_to_data ../barcelona I'm getting the error - ImportError: No module named soccer3d

Currently FIFA can only be launched from Origin, how to launch it by RenderDoc in order to save captures?

Hi,
I am using your code to test the AR/VR. Thanks for your contribution. I need one help in executing the code. I am stuck at last stage "demo/simple_visualization.py" script. I have successfully completed all the steps and data is been created. But when I run demo/simple_visualization.py I am getting the below error.

[i] Requesting "texture.png" from remote server
I0122 12:15:35.454744 13646 init.py:49] [i] Requesting "texture.png" from remote server
[w] Data not available on remote server
W0122 12:15:36.370683 13646 init.py:53] [w] Data not available on remote server
Traceback (most recent call last):
File "demo/simple_visualization.py", line 157, in
field['u_texture'] = data.get(join(opt.path_to_data, 'texture.png'))
File "/usr/local/lib/python3.6/dist-packages/glumpy/data/init.py", line 167, in get
return load(filename, *args, **kwargs)
File "/usr/local/lib/python3.6/dist-packages/glumpy/data/init.py", line 173, in load
extension = os.path.basename(filename).split('.')[-1]
File "/usr/lib/python3.6/posixpath.py", line 146, in basename
p = os.fspath(p)
TypeError: expected str, bytes or os.PathLike object, not NoneType

Please guide me to resolve this issue.

I want to try just the camera calibration (on one image) to check how it works. Can you please tell me, what is coarse_mask?

As stated here coarse_mask = self.get_mask_from_detectron(0) it's some mask from Detectron? But Detectron is used here for human detections?

P.S. Actually I tried to build the whole project but failed - too much dependencies, conflicts, etc. So now I'm trying to use only calibration part. But without Detectron I just can't check what is this coarse_mask Thanks.

When I executed refining instance segmentation, I got the error.

./soccer3d/instancesegm/instancesegm: No such file or directory

The command line is

./soccer3d/instancesegm/instancesegm --path_to_data $DATADIR/players/ --thresh 1.5 --path_to_model ./soccer3d/instancesegm/model.yml.gz

I checked this directory. There is no instancesegm file under soccer3d/instancesegm directory.

I think the author might forget to compile, or forget to tell us.
The solution is to go to directory soccer3d/instancesegm/, and execute
cmake .
make

Hi @krematas ,
Thanks for this amazing repo.
I successfully executed the first step which generated 146 files in

$DATADIR/detectron

.
When I'm running demo/caliber_video.py, It's showing a figure and still running long time.
When I cut the shown figure,the code is throwing

IndexError

.Then I run the file again and I could find the code line
db.calibrate_camera() in caliber_video.py is running.
youtube.txt, metadata/detectron.p and an empty folder calib are created.
Some files are still to be created.
Can we make it little bit faster somehow?
Thanks in advance.

Hello , could you update the project ...caffee2 now is emigrated into PyTorch ( can't run with Anaconda ) and I can't figure out how to update it.
also , \DETECTRON2 is released which comes with enhancement it would be better to document the process more clearly .(specially the cloud points to unity part )
if possible use only one library to do all tasks preferably using a NeRf for better consistently for posse estimations on onnx model notebook.

thanks ,

Hi @krematas ,
I'm getting Segmentation fault (core dumped) while running following step:-

python3 soccer3d/soccerdepth/test.py --path_to_data $DATADIR/players --modelpath $MODELPATH

Where MODELPATH=/home/ujjawal/my_work/soccerontable/MODEL_maskRCNN/model.pth
DATADIR=/home/ujjawal/my_work/soccerontable/barcelona

Did anyone face this issue ?
Please suggest me some ways to handle this.

It's trouble to build opencv and opencv contrib from source. There are hundreds of errors during compilation. I found a much easier way.

Actually, opencv in c++ is needed. So pip to install is not feasible.

I was wondering, as I didn't see any images from top-down what the result would look like. Btw. do I need the holo-lens to test that?

$ python2 tools/infer_subimages.py --cfg configs/12_2017_baselines/e2e_mask_rcnn_R-101-FPN_2x.yaml --output-dir $DATADIR/detectron --image-ext jpg --wts https://s3-us-west-2.amazonaws.com/detectron/35861858/12_2017_baselines/e2e_mask_rcnn_R-101-FPN_2x.yaml.02_32_51.SgT4y1cO/output/train/coco_2014_train:coco_2014_valminusminival/generalized_rcnn/model_final.pkl $DATADIR/images/

Traceback (most recent call last):
File "tools/infer_subimages.py", line 219, in
main(args)
File "tools/infer_subimages.py", line 103, in main
model = infer_engine.initialize_model_from_cfg()
TypeError: initialize_model_from_cfg() takes at least 1 argument (0 given)

I checked the "test_engin.py" file in Detectron/detectron/tools and found the function initialize_model_from_cfg needs at least 1 weight_file argument

def initialize_model_from_cfg(weights_file, gpu_id=0):
"""Initialize a model from the global cfg. Loads test-time weights and
creates the networks in the Caffe2 workspace.
"""
model = model_builder.create(cfg.MODEL.TYPE, train=False, gpu_id=gpu_id)
net_utils.initialize_gpu_from_weights_file(
model, weights_file, gpu_id=gpu_id,
)
model_builder.add_inference_inputs(model)
workspace.CreateNet(model.net)
workspace.CreateNet(model.conv_body_net)
if cfg.MODEL.MASK_ON:
workspace.CreateNet(model.mask_net)
if cfg.MODEL.KEYPOINTS_ON:
workspace.CreateNet(model.keypoint_net)
return model