Into the Fog: Evaluating Robustness of Multiple Object Tracking
This addresses the problem of unreliable MOT in real-world scenarios with fog for autonomous driving and surveillance, but it is incremental as it focuses on benchmarking rather than proposing new tracking methods.
The paper tackled the lack of robustness in Multiple Object Tracking (MOT) methods under adverse atmospheric conditions like fog by introducing a physics-based volumetric fog simulation method for arbitrary datasets, and it revealed limitations of state-of-the-art trackers when evaluated on a fog-augmented benchmark.
State-of-the-art Multiple Object Tracking (MOT) approaches have shown remarkable performance when trained and evaluated on current benchmarks. However, these benchmarks primarily consist of clear weather scenarios, overlooking adverse atmospheric conditions such as fog, haze, smoke and dust. As a result, the robustness of trackers against these challenging conditions remains underexplored. To address this gap, we introduce physics-based volumetric fog simulation method for arbitrary MOT datasets, utilizing frame-by-frame monocular depth estimation and a fog formation optical model. We enhance our simulation by rendering both homogeneous and heterogeneous fog and propose to use the dark channel prior method to estimate atmospheric light, showing promising results even in night and indoor scenes. We present the leading benchmark MOTChallenge (third release) augmented with fog (smoke for indoor scenes) of various intensities and conduct a comprehensive evaluation of MOT methods, revealing their limitations under fog and fog-like challenges.