ARTHuS: Adaptive real-time human segmentation in sports through online distillation
(2019) IEEE/CVF International conference on Computer Vision and Pattern Recognition, workshop on Computer Vision in Sports (16.June.2019)
Files
paper_camera_ready-compressed.pdf
Preprint, compressed
Open Access - Adobe PDF
- 852.73 KB
Details
- Authors
- Abstract
- Semantic segmentation can be regarded as a useful tool for global scene understanding in many areas, including sports, but has inherent difficulties, such as the need for pixel-wise annotated training data and the absence of well-performing real-time universal algorithms. To alleviate these issues, we sacrifice universality by developing a general method, named ARTHuS, that produces adaptive real-time game-specific networks for human segmentation in sports videos, without requiring any manual annotation. This is done by an online knowledge distillation process, in which a fast student network is trained to mimic the output of an existing slow but effective universal teacher network, while being periodically updated to adjust to the latest play conditions. As a result, ARTHuS allows to build highly effective real-time human segmentation networks that evolve through the match and that sometimes outperform their teacher. The usefulness of producing adaptive game-specific networks and their excellent performances are demonstrated quantitatively and qualitatively for soccer and basketball games.
- Affiliations
Citations
Cioppa, A., Deliège, A., Istasse, M., De Vleeschouwer, C., & Van Droogenbroeck, M. (2019). ARTHuS: Adaptive real-time human segmentation in sports through online distillation. Proceedings of IEEE/CVF International conference on Computer Vision and Pattern Recognition, 1(1), 10. https://hdl.handle.net/2078.5/60711 (Original work published 2019)