What is the base-model of "pose_estimation.onnx"?

Dear NVIDIA Developers,
I have a simple question. Would you answer this?
What is the base model of pose_estimation.onnx?

I installed it from NVIDIA-AI-IOT/ deepstream_pose_estimation.

I confirmed its structure, but it seems that it is different it from resnet18_baseline_att_224x224_A.json

Kind regards

Hi @Kazukie ,
The model is from GitHub - NVIDIA-AI-IOT/trt_pose: Real-time pose estimation accelerated with NVIDIA TensorRT

Hi @mchi , thank you for your quick reply.
What I pointed out is the pose_estimation.onnx(deepstream_pose_estimation/pose_estimation.onnx at master · NVIDIA-AI-IOT/deepstream_pose_estimation · GitHub) is not based with resnet18_baseline_att_224x224_A_epoch_249.pth - Google Drive you mentioned.

I compared both onnx models. The resnet18_baseline_att_224x224_A_epoch_249.pth was converted by export_for_isaac.py .

I visualized the network structure on this site(https://netron.app/) and compared them.
I guessed that their network structures would be the same if pose_estimation.onnx model was based on resnet18_baseline_att_244x244_epoch_249 model. But it was different, so I opened this topic.

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

Sorry for long delay!
Is this still needed?


This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.