This project implements a pedestrian detection and segmentation model using Near-Infrared (NIR) and Depth (D) images through a Refined Attention Feature-enhanced Fusion (RAFF) architecture.