Self-supervised part segmentation via motion imitation

作者：

Highlights：

• A self-supervised part segmentation via motion imitation is proposed.

• The complementarity of key point information and part information is investigated.

• The proposed method can produce more semantically consistent and meaningful parts than SOTA methods.

• The effectivity and validity of the proposed method have verified through extensive experiments.

摘要

Highlights•A self-supervised part segmentation via motion imitation is proposed.•The complementarity of key point information and part information is investigated.•The proposed method can produce more semantically consistent and meaningful parts than SOTA methods.•The effectivity and validity of the proposed method have verified through extensive experiments.

论文关键词：Motion imitation,Self-supervised,Part segmentation

论文评审过程：Received 14 May 2021, Revised 16 November 2021, Accepted 18 January 2022, Available online 26 January 2022, Version of Record 10 February 2022.

论文官网地址：https://doi.org/10.1016/j.imavis.2022.104393