Tilting at windmills

Data augmentation for deep pose estimation does not help with occlusions

Master Thesis (2020)
Author(s)

R.J. Pytel (TU Delft - Electrical Engineering, Mathematics and Computer Science)

Contributor(s)

Jan van Gemert – Mentor (TU Delft - Pattern Recognition and Bioinformatics)

O.S. Kayhan – Mentor (TU Delft - Pattern Recognition and Bioinformatics)

Marcel Reinders – Graduation committee member (TU Delft - Pattern Recognition and Bioinformatics)

Cynthia CS Liem – Graduation committee member (TU Delft - Multimedia Computing)

Faculty
Electrical Engineering, Mathematics and Computer Science
Copyright
© 2020 Rafal Pytel
More Info
expand_more
Publication Year
2020
Language
English
Copyright
© 2020 Rafal Pytel
Graduation Date
31-08-2020
Awarding Institution
Delft University of Technology
Faculty
Electrical Engineering, Mathematics and Computer Science
Reuse Rights

Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.

Abstract

Occlusion degrades the performance of human pose estimation. In this paper, we introduce targeted keypoint and body part occlusion attacks. The effects of the attacks are systematically analyzed on the best-performing methods. In addition, we propose occlusion specific data augmentation techniques against keypoint and part attacks. Our extensive experiments show that human pose estimation methods are not robust to occlusion and data augmentation does not solve the occlusion problems.

Files

MSc_Thesis_Final.pdf
(pdf | 22 Mb)
- Embargo expired in 24-08-2020
License info not available