Duncan Zauss, Sven Kreiss, Alexandre Alahi
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.
| Task | Dataset | Metric | Value | Model |
|---|---|---|---|---|
| Pose Estimation | ApolloCar3D | Detection Rate | 91.9 | Zauss et al. |
| 3D | ApolloCar3D | Detection Rate | 91.9 | Zauss et al. |
| 2D Human Pose Estimation | COCO-WholeBody | WB | 60.4 | Zauss et al. |
| 2D Human Pose Estimation | COCO-WholeBody | body | 69.6 | Zauss et al. |
| 2D Human Pose Estimation | COCO-WholeBody | face | 85 | Zauss et al. |
| 2D Human Pose Estimation | COCO-WholeBody | foot | 63.4 | Zauss et al. |
| 2D Human Pose Estimation | COCO-WholeBody | hand | 52.9 | Zauss et al. |
| 1 Image, 2*2 Stitchi | ApolloCar3D | Detection Rate | 91.9 | Zauss et al. |