Conversion of model weights for human pose estimation model to ONNX results in nonsensical pose estimation

Accelerated Computing Intelligent Video Analytics DeepStream SDK
, ,
13

Hi,

Thanks for your patience.
Please give this script a try: export_for_isaac.py (8.4 KB)

$ python3 export_for_isaac.py --input_checkpoint resnet18_baseline_att_224x224_A_epoch_249.pth --input_topology ../../tasks/human_pose/human_pose.json

We can get correct result with the ONNX file generated from the above script.

Thanks.

2 Likes