documentation:vision:pose_estimation

Differences

This shows you the differences between two versions of the page.

Link to this comparison view

Next revision		 Previous revision
documentation:vision:pose_estimation [2022-03-22 11:03] – created Robbert Hofmandocumentation:vision:pose_estimation [2022-09-02 16:04] (current) – external edit 127.0.0.1
Line 4: Line 4:
   * https://github.com/DLR-RM/AugmentedAutoencoder   * https://github.com/DLR-RM/AugmentedAutoencoder
   * https://github.com/shbe-aau/multi-pose-estimation   * https://github.com/shbe-aau/multi-pose-estimation
 +  * https://github.com/HampusAstrom/yolov3 (reduced, simpler repo to get things working on heron)
 +  * https://github.com/robberthofmanfm/yolo/releases (used as a fileserver for the weights to avoid git LFS)
  
-The former two are inspirations to the latter, which is the one we use. The authors are  [[hampus.astrom@cs.lth.se|Hampus Astrom]] and [[shbe@create.aau.dk|Stefan Bengtson]].+The former two are inspirations to the third one. The authors are  [[hampus.astrom@cs.lth.se|Hampus Astrom]] and [[shbe@create.aau.dk|Stefan Bengtson]].
  
 It uses an autoencoder from pytorch3d, which itself is built on top of pytorch. It uses an autoencoder from pytorch3d, which itself is built on top of pytorch.
Line 11: Line 13:
 For training, the most important file is [[https://github.com/shbe-aau/multi-pose-estimation/blob/master/multi-pose/experiments/experiment_template.cfg|experiment_template.cfg]] For training, the most important file is [[https://github.com/shbe-aau/multi-pose-estimation/blob/master/multi-pose/experiments/experiment_template.cfg|experiment_template.cfg]]
  
 +==== Troubleshooting ====
 +Since it runs in a docker container and uses nvidia, you'll want to pass the ''--gpus all'' flag to ''docker run''. \\
 +However, when calling ''docker build'', no such flag exists. The build will fail. \\
 +For me, the answer was to specify the default docker runtime as one that uses nvidia-docker.\\
 +  - ''sudo apt-get install nvidia-container-runtime''  # follow this guide https://nvidia.github.io/nvidia-container-runtime/
 +  - edit ''/etc/docker/daemon.json'' and make sure it looks something like this: <code> {
 +    "runtimes": {
 +        "nvidia": {
 +            "path": "/usr/bin/nvidia-container-runtime",
 +            "runtimeArgs": []
 +         } 
 +    },
 +    "default-runtime": "nvidia" 
 +}
 +</code>
 +  - ''sudo systemctl restart docker''
 +I've learned this from: 
 +https://stackoverflow.com/questions/59691207/docker-build-with-nvidia-runtime
  • documentation/vision/pose_estimation.1647943404.txt.gz
  • Last modified: 2022-09-02 16:04
  • (external edit)