inspose

InsPose: Instance-Aware Networks for Single-Stage Multi-Person Pose Estimation

Results and Models

Model	Backbone	Lr schd	mAP	AP50	AP75	APM	APL	Config	Download
InsPose	R-50	3x	65.2	87.2	71.3	60.6	72.2	config	Google Drive | BaiduYun

NOTE

1. The training annotation is generated by tools\dataset_converters\keypoint2pseudo_box.py

Citation

@inproceedings{shi2021inspose,
  title={Inspose: instance-aware networks for single-stage multi-person pose estimation},
  author={Shi, Dahu and Wei, Xing and Yu, Xiaodong and Tan, Wenming and Ren, Ye and Pu, Shiliang},
  booktitle={Proceedings of the 29th ACM International Conference on Multimedia},
  pages={3079--3087},
  year={2021}
}