3D hand pose and shape estimation from RGB images for keypoint-based hand gesture recognition

作者：

Highlights：

• presentation of a comprehensive end-to-end keypoint framework for 3D hand pose/shape estimation.

• design of a multi-task feature extractor, optimized viewpoint encoder, and re-projection procedure for stable outputs.

• evaluation on a second task, i.e., hand-gesture recognition, where other keypoint-based approaches are outperformed.

摘要

•presentation of a comprehensive end-to-end keypoint framework for 3D hand pose/shape estimation.•design of a multi-task feature extractor, optimized viewpoint encoder, and re-projection procedure for stable outputs.•evaluation on a second task, i.e., hand-gesture recognition, where other keypoint-based approaches are outperformed.

论文关键词：Hand pose estimation,Hand shape estimation,Deep learning,Hand gesture recognition

论文评审过程：Received 29 September 2021, Revised 5 April 2022, Accepted 29 April 2022, Available online 30 April 2022, Version of Record 5 May 2022.

论文官网地址：https://doi.org/10.1016/j.patcog.2022.108762