Joint COCO and LVIS workshop at ECCV 2020: COCO Keypoint Challenge Track Technical Report: UDP++

semanticscholar(2020)

引用 4|浏览2
暂无评分
摘要
In this report, we present our human pose estimation system, UDP++. UDP++ takes deep HRNet as network backbone and is equipped with Unbiased Data Processing and Occlusion Augmentation supervision. On the COCO-2020 keypoint test-dev dataset, the UDP++ system achieves 80.8 AP, which surpasses the 2019 winning result by 1.6 AP. The source code will be publicly available for further research in https://github.com/HuangJunJie2017/UDP-Pose.
更多
查看译文
AI 理解论文
溯源树
样例
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要