Oriane Siméoni to detail DINOv3 and the case for scaling vision self-supervised learning at CVPR 2026 · Digg