Keypoint Communities

TL;DR

This paper introduces a fast bottom-up approach for keypoint detection that models poses as graphs, using community detection insights to improve accuracy in human and object pose estimation with over 100 keypoints.

Contribution

It proposes a novel graph-based method leveraging community detection and centrality measures to enhance keypoint detection and pose estimation accuracy.

Findings

Outperforms previous methods on human pose estimation with 133 keypoints.

Effectively generalizes to car pose estimation.

Uses graph centrality to weight training of keypoints.

Abstract

We present a fast bottom-up method that jointly detects over 100 keypoints on humans or objects, also referred to as human/object pose estimation. We model all keypoints belonging to a human or an object -- the pose -- as a graph and leverage insights from community detection to quantify the independence of keypoints. We use a graph centrality measure to assign training weights to different parts of a pose. Our proposed measure quantifies how tightly a keypoint is connected to its neighborhood. Our experiments show that our method outperforms all previous methods for human pose estimation with fine-grained keypoint annotations on the face, the hands and the feet with a total of 133 keypoints. We also show that our method generalizes to car poses.