This project uses the DINOv3 (self-distillation with no labels v3) model for real-time object tracking. DINOv3 is a state-of-the-art computer vision model developed by Meta AI. It is trained using self-supervised learning, meaning it learns rich visual features from images without requiring any manual annotations.