Training PoseClassificationNet for customised activities

[edit_or]

• Network Type (Detectnet_v2/Faster_rcnn/Yolo_v4/LPRnet/Mask_rcnn/Classification/etc) PoseClassificationNet

I am training PoseClassificationNet for customized activities.
I am not using all 34 key points.
The features used in training are speed and length converted from some keypoints.
How can I use PoseClassificationNet to train customized dataset?
Because PoseClassificationNet needs to specify graph_layout and graph_strategy in training.
Is it possible to train using PoseClassificationNet?

[Morganh]

@edit_or

I am not using all 34 key points.

You can refer to tao_pytorch_backend/nvidia_tao_pytorch/cv/pose_classification/model/st_gcn.py at 9c2d94c0635b1117edfea85a94a6e3d0ead53754 · NVIDIA/tao_pytorch_backend · GitHub and modify it to meet your key points.
Similar topic is PoseClassificationNet Model training on custom dataset.

[edit_or]

This post has only question and no solution

https://forums.developer.nvidia.com/t/poseclassificationnet-model-training-on-custom-dataset/272886

[Morganh]

Different keypoints are expected to work. There are some dataset mentioned in tao_pytorch_backend/nvidia_tao_pytorch/cv/pose_classification/model/st_gcn.py at 9c2d94c0635b1117edfea85a94a6e3d0ead53754 · NVIDIA/tao_pytorch_backend · GitHub. They have different keypoints.

[edit_or]

Thank you. Let me try

[yingliu]

There is no update from you for a period, assuming this is not an issue anymore. Hence we are closing this topic. If need further support, please open a new one. Thanks