Temporal Representation Learning on Monocular Videos for 3D Human Pose Estimation

• Published in: IEEE transactions on pattern analysis and machine intelligence Vol. 45; no. 5; pp. 6415 - 6427
• Main Authors: Honari, Sina, Constantin, Victor, Rhodin, Helge, Salzmann, Mathieu, Fua, Pascal
• Format: Journal Article
• Language: English
• Published: United States IEEE 01.05.2023; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: 3D human pose; Algorithms; Cameras; contrastive learning; Crops; Error reduction; Feature extraction; Humans; Image reconstruction; Learning; Pose estimation; Temporal feature extraction; Three-dimensional displays; Unsupervised learning; unsupervised representation learning; Video; Videotape Recording

In this article we propose an unsupervised feature extraction method to capture temporal information on monocular videos, where we detect and encode subject of interest in each frame and leverage contrastive self-supervised (CSS) learning to extract rich latent vectors. Instead of simply treating th...