Yongtao Ge

Hi, @gs-ren thanks for your interest in our paper. the uncertainty estimation is implemented [here](https://github.com/aim-uofa/Poseur/blob/9358ed54dc597f9c46b114e8f57b018b5af55ddd/mmpose/models/detectors/poseur.py#L227)

It is just a hyperparameter that scales the keypoint score, we set it to 0.7 by default.

1. If the original video filename contains `' '` , the cap.read() will return false. 2. In this case, the generated image name is weird.

I don’t have the correct version either. could you give a link from which you borrow the code? The total number of generated images are not equal 417772.

@Arthur151 请问一下谷歌网盘的h36m数据是通过这个处理代码得到的吗 我尝试这个代码得到的图片数量不对

I use this code to train the whole dataset with basic transformer model. The training process seems correct as the loss keeps decreasing. But I found that the model can't...

yep! some modules work fine while others not. I could not locate the bug.

I just come across the error"cudaSuccess (2 vs. 0) out of memory". My GPU is GTX850 and I use the code to predict only one image. Is there something wrong?

Thank you! I would have a try.

We use 8 gpus for training for our experiments. However, if you do not have enough resources, you can use fewer gpus (adjust learning rate and batchsize, accordingly).