task:Human pose estimation
motivation:regressing keypoint positions accurately needs to learn representations that focus on the keypoint regions.
background:
two main paradigms: top-down and bottom-up.
top-down: first detects the person and then performs single-person pose estimation for each detected person.
bottom-up: either directly regresses the keypoint positions belonging to the same person, or detects and groups the keypoints.
The top-down paradigm is more accurate but more costly due to an extra person detection pro- cess, and the bottom-up paradigm, the interest of this paper, is more efficient. 但是需要繁重的后处理工作。
method:
两个创新点:
1).自适应卷积激活关键点周围像素的区域以学习到新特征
2).multi-branch parallel adaptive convolutions to learn disentangled representations for the regression of the K keypoints, so that each representation focuses on the cor- responding keypoint region.
Disentangled Keypoint Regression
自适应卷积
Separate regression.