search

MiracleDance / PoseRAC

PoseRAC: Pose Saliency Transformer for Repetitive Action Counting
MIT License
13 stars 2 forks source link

Cannot replicate the results #9

Closed wanghao-cst closed 1 year ago

wanghao-cst commented 1 year ago

Hello,

I recently downloaded and attempted to replicate the results of your open-source code. However, I'm facing challenges in achieving the reported accuracy as mentioned by the author in the documentation.

Here are the steps I followed: step0: download the RepCount LLSP dataset step1: generate pose for test set python pre_test.py --config ./RepCount_pose_config.yaml step2: evaluate on the model you provided python eval.py --config ./RepCount_pose_config.yaml --ckpt ./best_weights_PoseRAC.pth

image

The MAE and OBO are 0.243 and 0.546 respectively.

Thank you for your attention to this matter. I look forward to your response.

MiracleDance commented 1 year ago

I try your steps again, but the results are still the same with that in README.md. I'm not sure whether you have other different operations.