Comments (13)
Thanks for your interest! In this project we don't directly take .bin files as inputs. However, I have a naive and incomplete python script for converting the .bin files to .png via direct projection. Note that the projected depth maps from LiDaR pointclouds are not exactly the same as those in KITTI depth completion dataset as Uhrig et al. [3DV2017] composes ego motion compensation. I hope this script may help.
bin2png.txt
from penet_icra2021.
hi, @JUGGHM I tried the code what you gave for converting .bin to .png format, but I have some doubt like, why still it need to read depth from .png because this is the code to read depth .bin file and projecting in-camera coordinates and save that projection in .png file right? I would like to convert .bin to .png for depth completion in the custom dataset. Thank you very much
from penet_icra2021.
hi, @JUGGHM I tried the code what you gave for converting .bin to .png format, but I have some doubt like, why still it need to read depth from .png because this is the code to read depth .bin file and projecting in-camera coordinates and save that projection in .png file right? I would like to convert .bin to .png for depth completion in the custom dataset. Thank you very much
I just used the depth file to ensure whether the projected result is right, and you could read no depth maps actually. I didn't involve it in PENet but I think you could modify this incomplete script for your own use. The pro_depth
variable is the projected sparse depth in numpy format and I think it is what you want. And you might initialize it manually but not from #pro_depth = np.zeros_like(dc_depth)
.
from penet_icra2021.
Yes Exactly, because currently I am having .npy format of depth map for 3D object detection. Now can i use this .npy file in your code to estimate dense depth map and i have respective Images for those depth also but i am little confusion about calibration file because the calibration file of depth completion datasets is different from 3D object detection ! can you help me to solve this problem ?
from penet_icra2021.
Yes Exactly, because currently I am having .npy format of depth map for 3D object detection. Now can i use this .npy file in your code to estimate dense depth map and i have respective Images for those depth also but i am little confusion about calibration file because the calibration file of depth completion datasets is different from 3D object detection ! can you help me to solve this problem ?
In my script I used the calibration matrix of KITTI raw dataset, and I am not quite familiar with 3D object detection task.
from penet_icra2021.
oh okay thank you very much for your quick replies. can we change hyper parameter values in Loss functions while training ? because it mentioned (0,0,0) I am just curious will it work if i change it ?
from penet_icra2021.
oh okay thank you very much for your quick replies. can we change hyper parameter values in Loss functions while training ? because it mentioned (0,0,0) I am just curious will it work if i change it ?
It will work but the performance will change I think. It corresponds to the description of intermediate supervision in the implementation details in the paper.
from penet_icra2021.
oh okay thank you very much for your quick replies. can we change hyper parameter values in Loss functions while training ? because it mentioned (0,0,0) I am just curious will it work if i change it ?
I reviewed the code but found only loss items with 0, 0, 0 initialization. It is just for creating the variables.
from penet_icra2021.
oh maybe i maybe misunderstood the logic over there, then what value you assigned as hyper parameter for your loss functions ? Thank you!
from penet_icra2021.
oh maybe i maybe misunderstood the logic over there, then what value you assigned as hyper parameter for your loss functions ? Thank you!
w_st1
and w_st2
from penet_icra2021.
Can you explain more about this hyper parameter ? Thank you
from penet_icra2021.
oh okay thank you very much for your quick replies. can we change hyper parameter values in Loss functions while training ? because it mentioned (0,0,0) I am just curious will it work if i change it ?
It will work but the performance will change I think. It corresponds to the description of intermediate supervision in the implementation details in the paper.
from penet_icra2021.
@Laihu08
Have you solved this problem using the author's script?
I want to convert KITTI object dataset into depth too.
from penet_icra2021.
Related Issues (20)
- How to use the sparse depth?
- How to use the sparse depth? HOT 4
- Modify the backbone network HOT 3
- lightweight deployment of PENet network
- How to infer PENet for KITTI object task? HOT 3
- broken PNG file HOT 1
- the difference intrinsic parameters between train and test HOT 4
- Modify DA-CSPN++ for single branch input HOT 1
- some questions about the implement of CSPN HOT 4
- Model mismatch at inference time HOT 1
- Tensor Dimension Mismatch HOT 1
- How many parameters dose PENet have? HOT 2
- About training HOT 1
- How to implement the model with ROS?
- can't download the pretrained PENet Model
- Pretrained on NYU dataset?
- RuntimeError: CUDA out of memory.
- confindence map
- lower lidar scanline input
- Is there a trained model instead of a pre-trained model
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from penet_icra2021.