Occlusion Geodesics for Online Multi-Object Tracking

Robust multi-object tracking-by-detection requires the correct assignment of noisy detection results to object tra-jectories. We address this problem by proposing an on-line approach based on the observation that object detec-tors primarily fail if objects are significantly occluded. In contrast to most existing work, we only rely on geometric information to efficiently overcome detection failures. In particular, we exploit the spatio-temporal evolution of occlusion regions, detector reliability, and target motion prediction to robustly handle missed detections. In com-bination with a conservative association scheme for visi-ble objects, this allows for real-time tracking of multiple objects from a single static camera, even in complex sce-narios. Our evaluations on publicly available multi-object tracking benchmark datasets demonstrate favorable perfor-mance compared to the state-of-the-art in online and offline multi-object tracking. 1