Yining Li

Need a more informative PR title.

[Labelbee](https://github.com/open-mmlab/labelbee-client) is what you need. [LabelMe](https://github.com/wkentaro/labelme) is also a popular tool for keypoint annotation.

Both of them support multiple annotation types like bbox (with class label), keypoint, and attribute.

Training of SmoothNet has not been supported in MMPose yet. Please refer to the official repo https://github.com/cure-lab/SmoothNet

1. Warm-up is a commonly used training strategy that gradually increases the learning rate at the beginning of the training, instead of directly starting with a large learning rate. 2....

1. If `lr=5e-4, warmup_ratio=0.001, warmup_iters=500`, it means the learning rate will increase from `5e-4 * 0.001` to `5e-4` during the first 500 epoch. Please check the code in mmcv for...

1. The warm-up will be applied if the argument `warmup` is set. It will use a default warm-up ratio if the argument `warmup_ratio` is missing. Please check the code in...

@wusize Could you please rebase the PR branch to `dev-0.28`? And the CI failed and may need a check.

@wusize Could you please modify `mmpose/__init__.py` and set `maximum_mmcv_version=1.7.0` to avoid the CI failure?

@wusize The mmcv version issue has been fixed in the master branch. But it's better not to rebase onto the master now since it may cause confusion when merging dev-0.29...