Yining Li

Thanks for your feedback. We will consider providing 2D SmoothNet pre-trained models. Although, COCO may not be suitable since training SmoothNet requires keypoint annotations on frame sequences.

> 3D Human Mesh [frankmocap](https://github.com/facebookresearch/frankmocap) Thanks for your feedback. 3D human mesh recovery is no longer supported in MMPose. We have [MMHuman3D](https://github.com/open-mmlab/mmhuman3d) for this task and you are welcome to...

@maowayne123 Thanks for the suggestion. It looks great. Would you like to create a PR to support this feature in MMPose? I think maybe we can also add an argument...

@Annbless Thank you very much for your help! It's inspiring to see that MMPose is facilitating great research works like ViTPose and bringing convenience to the community. And the contribution...

Human3.6M dataset has been used for 2 different tasks in MMPose, namely 3D keypoint detection and mesh recovery, with different annotation structures and preparation processes. For 3D keypoint detection, the...

I am not sure why there is such a large difference between the data sizes. Maybe it's because of the video2image approach? We use OpenCV while [CHUNYUWANG/H36M-Toolbox](https://github.com/CHUNYUWANG/H36M-Toolbox) directly uses FFmpeg...

> Human3.6M dataset has been used for 2 different tasks in MMPose, namely 3D keypoint detection and mesh recovery, with different annotation structures and preparation processes. For 3D keypoint detection,...

> > I am not sure why there is such a large difference between the data sizes. Maybe it's because of the video2image approach? We use OpenCV while [CHUNYUWANG/H36M-Toolbox](https://github.com/CHUNYUWANG/H36M-Toolbox) directly...

@Indigo6 BTW, would you be interested in an internship in OpenMMLab? If so please reach me via [email protected] :)

There were two reasons: 1) The developer was an intern in the mmpose team and he left for an exchange opportunity before these PRs were ready to merge; 2) Coarse-to-fine...