search

fire717 / movenet.pytorch

A Pytorch implementation of MoveNet from Google. Include training code and pre-trained model.
MIT License
370 stars 87 forks source link

Can you give a good weight #47

Open yuqilol opened 1 year ago

yuqilol commented 1 year ago

hey，i wanna you great pth like e100_valacc0.98349.pth，please help me。

rishabh063 commented 7 months ago

its present in the output folder named output/e118_valacc0.79805.pth

dhillyon commented 5 months ago

Hi! Did you have any luck with this? I can't seem to get accuracy above 79.5% in training.

fire717 commented 5 months ago

Thank you for your interest in this repo. I apologize that I am unable to provide the model at this time as it was trained on the intranet during my previous job at the company. And that is based on our business data, with more accurate labeling and a relatively simple validation set, so the score is relatively high. I don't know what your specific scenario is. If it's optimization for academic paper growth points, then movenet is not a good baseline because it has not published any papers or has no official open source implementation. If it is for your own project, it is recommended that you supplement more training data. At the same time, if it is not limited to using movenet, it is now recommended that you try the newly released RTMPose.