Quo Vadis: Is Trajectory Forecasting the Key Towards Long-Term Multi-Object Tracking?
This addresses a key limitation in multi-object tracking for applications like autonomous driving, where long occlusions hinder reliable tracking, representing an incremental but impactful advance.
The paper tackles the problem of bridging long occlusion gaps in multi-object tracking by using trajectory forecasting to reduce the search space for associations, improving state-of-the-art trackers on the MOTChallenge dataset with significant gains in long-term performance.
Recent developments in monocular multi-object tracking have been very successful in tracking visible objects and bridging short occlusion gaps, mainly relying on data-driven appearance models. While we have significantly advanced short-term tracking performance, bridging longer occlusion gaps remains elusive: state-of-the-art object trackers only bridge less than 10% of occlusions longer than three seconds. We suggest that the missing key is reasoning about future trajectories over a longer time horizon. Intuitively, the longer the occlusion gap, the larger the search space for possible associations. In this paper, we show that even a small yet diverse set of trajectory predictions for moving agents will significantly reduce this search space and thus improve long-term tracking robustness. Our experiments suggest that the crucial components of our approach are reasoning in a bird's-eye view space and generating a small yet diverse set of forecasts while accounting for their localization uncertainty. This way, we can advance state-of-the-art trackers on the MOTChallenge dataset and significantly improve their long-term tracking performance. This paper's source code and experimental data are available at https://github.com/dendorferpatrick/QuoVadis.