Key Frame Proposal Network for Efficient Pose Estimation in Videos
arXiv:2007.15217
Abstract
Human pose estimation in video relies on local information by either estimating each frame independently or tracking poses across frames. In this paper, we propose a novel method combining local approaches with global context. We introduce a light weighted, unsupervised, key frame proposal network (K-FPN) to select informative frames and a learned dictionary to recover the entire pose sequence from these frames. The K-FPN speeds up the pose estimation and provides robustness to bad frames with occlusion, motion blur, and illumination changes, while the learned dictionary provides global dynamic context. Experiments on Penn Action and sub-JHMDB datasets show that the proposed method achieves state-of-the-art accuracy, with substantial speed-up.
arXiv commentAccepted by ECCV2020
References in corpus (9)
- Two-Stream Convolutional Networks for Action Recognition in Videos
- Deep Residual Learning for Image Recognition
- DeepPose: Human Pose Estimation via Deep Neural Networks
- OpenPose: Realtime Multi-Person 2D Pose Estimation using Part Affinity Fields
- Stacked Hourglass Networks for Human Pose Estimation
- Feature Pyramid Networks for Object Detection
- Articulated Pose Estimation by a Graphical Model with Image Dependent Pairwise Relations
- Learning Feature Pyramids for Human Pose Estimation
- PoseTrack: Joint Multi-Person Pose Estimation and Tracking