Leveraging Language-Derived Appearance Knowledge to Enhance Pedestrian Detection
Leveraging the contextual understanding capabilities of large language models to formulate appearance knowledge elements and integrate them with visual cues can significantly enhance pedestrian detection performance.