A review of deep learning techniques for 2D and 3D human pose estimation

作者：

Highlights：

• Review up-to-date deep learning-based methods for 2D and 3D human pose estimation.

• Highlight the single-person and multi-person human pose estimation challenges

• Classify existing approaches into various categories based on their general architectures.

• Provide a discussion that summarizes the strengths and weaknesses of previous research.

• Point out possible future research directions.

摘要

•Review up-to-date deep learning-based methods for 2D and 3D human pose estimation.•Highlight the single-person and multi-person human pose estimation challenges•Classify existing approaches into various categories based on their general architectures.•Provide a discussion that summarizes the strengths and weaknesses of previous research.•Point out possible future research directions.

论文关键词：2D and 3D human pose estimation,Single-person and multi-person pose estimation,Deep learning,CNN,Computer vision

论文评审过程：Received 28 July 2021, Accepted 12 August 2021, Available online 18 August 2021, Version of Record 26 August 2021.

论文官网地址：https://doi.org/10.1016/j.imavis.2021.104282