Self-Supervised Learning via Conditional Motion Propagation
CVPR, pp. 1881-1889, 2019.
Intelligent agent naturally learns from motion. Various self-supervised algorithms have leveraged motion cues to learn effective visual representations. The hurdle here is that motion is both ambiguous and complex, rendering previous works either suffer from degraded learning efficacy, or resort to strong assumptions on object motions. In...More
PPT (Upload PPT)