MOTOR: A Multimodal Dataset for Two-Wheeler Rider Behavior Understanding

Researchers introduced MOTOR, a first-of-its-kind multimodal dataset for studying two-wheeler rider behavior in dense traffic.

The dataset includes synchronized multi-view video, eye-gaze, audio, and telemetry from 16 riders across 1,629 sequences.

It also provides labels for traffic context, maneuvers, and legality, making it useful for behavior and safety analysis.

Benchmark results show that RGB plus gaze and telemetry performs best for rider behavior understanding.

The work addresses a major data gap for motorcycles and scooters, especially relevant to road safety in the Global South.