A small monocular depth estimation model based on DINOv2 backbone, fine-tuned on the NYU Depth V2 dataset.