MAT: Motion-Aware Multi-Object Tracking

Shoudong Han,Piao Huang,Hongwei Wang,En Yu,Donghaisheng Liu,Xiaofeng Pan,Jun Zhao

DOI: https://doi.org/10.48550/arXiv.2009.04794

2020-09-18

Abstract:Modern multi-object tracking (MOT) systems usually model the trajectories by associating per-frame detections. However, when camera motion, fast motion, and occlusion challenges occur, it is difficult to ensure long-range tracking or even the tracklet purity, especially for small objects. Although re-identification is often employed, due to noisy partial-detections, similar appearance, and lack of temporal-spatial constraints, it is not only unreliable and time-consuming, but still cannot address the false negatives for occluded and blurred objects. In this paper, we propose an enhanced MOT paradigm, namely Motion-Aware Tracker (MAT), focusing more on various motion patterns of different objects. The rigid camera motion and nonrigid pedestrian motion are blended compatibly to form the integrated motion localization module. Meanwhile, we introduce the dynamic reconnection context module, which aims to balance the robustness of long-range motion-based reconnection, and includes the cyclic pseudo-observation updating strategy to smoothly fill in the tracking fragments caused by occlusion or blur. Additionally, the 3D integral image module is presented to efficiently cut useless track-detection association connections with temporal-spatial constraints. Extensive experiments on MOT16 and MOT17 challenging benchmarks demonstrate that our MAT approach can achieve the superior performance by a large margin with high efficiency, in contrast to other state-of-the-art trackers.

Computer Vision and Pattern Recognition

What problem does this paper attempt to address?

The paper attempts to address several key issues in multi-object tracking (MOT):

1. **Long-distance tracking in complex scenarios**: Existing detection-based multi-object tracking systems struggle to ensure the accuracy of long-distance tracking in the presence of challenges such as camera motion, fast movement, and occlusion, especially for small objects.

2. **Purity of trajectory fragments**: When occlusion or blurring occurs, existing methods tend to produce a large number of missed detections and incorrect associations, leading to impure trajectory fragments.

3. **Limitations of re-identification**: Although re-identification (ReID) technology is commonly used for long-distance matching or inter-frame association, due to partial detection noise, similar appearances, and lack of spatiotemporal constraints, ReID methods perform poorly in complex or low-resolution scenarios and are computationally expensive.

To address these issues, the paper proposes an enhanced multi-object tracking paradigm—**Motion-Aware Tracker (MAT)**, with the main innovations including:

- **Integrated Motion Localization (IML)**: Combining rigid camera motion and non-rigid pedestrian motion to form a unified motion localization module, improving the accuracy of motion prediction.
- **Dynamic Reconnection Context (DRC)**: Dynamically determining the reconnection window for different objects, balancing the robustness of long-distance tracking, and smoothing trajectory fragments caused by occlusion or blurring through a cyclic pseudo-observation update strategy.
- **3D Integral Image (3DII)**: Efficiently cutting off useless trajectory-detection association connections through spatiotemporal constraints, reducing computational costs.

These modules work together to enable MAT to perform significantly better than other advanced trackers in the MOT16 and MOT17 benchmarks.