Learning to Divide and Conquer for Online Multi-Target Tracking
arXiv:1509.03956
Abstract
Online Multiple Target Tracking (MTT) is often addressed within the tracking-by-detection paradigm. Detections are previously extracted independently in each frame and then objects trajectories are built by maximizing specifically designed coherence functions. Nevertheless, ambiguities arise in presence of occlusions or detection errors. In this paper we claim that the ambiguities in tracking could be solved by a selective use of the features, by working with more reliable features if possible and exploiting a deeper representation of the target only if necessary. To this end, we propose an online divide and conquer tracker for static camera scenes, which partitions the assignment problem in local subproblems and solves them by selectively choosing and combining the best features. The complete framework is cast as a structural learning task that unifies these phases and learns tracker parameters from examples. Experiments on two different datasets highlights a significant improvement of tracking performances (MOTA +10%) over the state of the art.
Cited by in corpus (8)
- Tracking the Trackers: An Analysis of the State of the Art in Multiple Object Tracking
- Real-Time Multiple Object Tracking - A Study on the Importance of Speed
- Fusion of Head and Full-Body Detectors for Multi-Object Tracking
- Knowledge Transfer for Scene-specific Motion Prediction
- Temporally Robust Global Motion Compensation by Keypoint-based Congealing
- Deep Hierarchical Machine: a Flexible Divide-and-Conquer Architecture
- Online Descriptor Enhancement via Self-Labelling Triplets for Visual Data Association
- Heuristic Search for Structural Constraints in Data Association