Online Multiperson Tracking-by-Detection from a Single, Uncalibrated Camera

Abstract

In this paper, we address the problem of automatically detecting and tracking a variable number of persons in complex scenes using a monocular, potentially moving, uncalibrated camera. We propose a novel approach for multiperson tracking-by-detection in a particle filtering framework. In addition to final high-confidence detections, our algorithm uses the continuous confidence of pedestrian detectors and online-trained, instance-specific classifiers as a graded observation model. Thus, generic object category knowledge is complemented by instance-specific information. The main contribution of this paper is to explore how these unreliable information sources can be used for robust multiperson tracking. The algorithm detects and tracks a large number of dynamically moving people in complex scenes with occlusions, does not rely on background modeling, requires no camera or ground plane calibration, and only makes use of information from the past. Hence, it imposes very few restrictions and is suitable for online applications. Our experiments show that the method yields good tracking performance in a large variety of highly dynamic scenarios, such as typical surveillance videos, webcam footage, or sports sequences. We demonstrate that our algorithm outperforms other methods that rely on additional information. Furthermore, we analyze the influence of different algorithm components on the robustness.

Tracking Interacting Objects Optimally Using Integer Programming

“…Because of their recursive nature, they are prone to errors that are difficult to recover from by using a post processing step. Particle-based approaches such as [13,24,8], among many others, partially address this issue by simultaneously exploring multiple hypotheses. However, they can handle only relatively small batches of temporal frames without their state space becoming unmanageably large, and often require careful parameter setting to converge.…”

Monocular Multiview Object Tracking with 3D Aspect Parts

“…Our approach falls in the category of tracking by detection methods [4,5,7,33,44], where category-level detectors are utilized to track the target of interest. However, in contrast to these methods, our focus is on tracking continuous 3D pose and 3D aspect parts.…”

Inter-Person Occlusion Handling with Social Interaction for Online Multi-Pedestrian Tracking

“…To this end, the optimized final-association score is capable of serving this purpose. In [13], a similar rationale has been used. However, our method differs from that study in terms of the occlusion modeling and dataassociation scheme.…”