Multi-Person Full Body Pose Estimation

Abstract

Multi-person pose estimation plays an important role in many fields. Although previous works have researched a lot on different parts of human pose estimation, full body pose estimation for multi-person still needs further research. Our work has developed an integrated model through knowledge distillation which can estimate full body poses. Trained based on the AlphaPose system and MSCOCO2017 dataset, our model achieves 51.5 mAP on the validation dataset annotated by ourselves.

Authors

Haoyi Zhu, Shanghai Jiao Tong University
Chen Jie, Shanghai Jiao Tong University
Shaofei Jiang, Shanghai Jiao Tong University

Code

Our source code is available on Github.

Paper

Our paper is available here.

Video

Our video is available on YouTube.

Examples of results