Self-supervised Multi-view Person Association and Its Applications
This addresses the challenge of tracking people in crowded, dynamic scenes for applications like multi-angle video production, though it is incremental as it builds on existing tracking-by-clustering methods.
The paper tackles the problem of markerless motion tracking of people in complex group activities from multiple moving cameras by developing a self-supervised framework to adapt a person appearance descriptor for reliable association across viewpoints and time, resulting in up to 18% improvement in association accuracy and 5 to 10 times more stable 3D skeleton tracking.
Reliable markerless motion tracking of people participating in a complex group activity from multiple moving cameras is challenging due to frequent occlusions, strong viewpoint and appearance variations, and asynchronous video streams. To solve this problem, reliable association of the same person across distant viewpoints and temporal instances is essential. We present a self-supervised framework to adapt a generic person appearance descriptor to the unlabeled videos by exploiting motion tracking, mutual exclusion constraints, and multi-view geometry. The adapted discriminative descriptor is used in a tracking-by-clustering formulation. We validate the effectiveness of our descriptor learning on WILDTRACK [14] and three new complex social scenes captured by multiple cameras with up to 60 people "in the wild". We report significant improvement in association accuracy (up to 18%) and stable and coherent 3D human skeleton tracking (5 to 10 times) over the baseline. Using the reconstructed 3D skeletons, we cut the input videos into a multi-angle video where the image of a specified person is shown from the best visible front-facing camera. Our algorithm detects inter-human occlusion to determine the camera switching moment while still maintaining the flow of the action well.