Human pose regression by combining indirect part detection and contextual information

• Published in: Computers & graphics Vol. 85; pp. 15 - 22
• Main Authors: Luvizon, Diogo C., Tabia, Hedi, Picard, David
• Format: Journal Article
• Language: English
• Published: Oxford Elsevier Ltd 01.12.2019; Elsevier Science Ltd; Elsevier
• Subjects: Computer Science; Computer vision; Datasets; Feature maps; Graphical representations; Ground truth; Human pose estimation; Image annotation; Man-machine interfaces; Neural and Evolutionary Computing; Neural nets; Regression; Signal and Image Processing; Source code; Neural nets; Computer vision; Human pose estimation

•A new human pose regression method from RGB images.•The proposed soft-argmax operation.•Contextual aggregation for refining predictions.•State-of-the-art results on important 2D pose estimation benchmarks among regression methods. [Display omitted] In this paper, we tackle the problem of human pose...