Romain BRÉGIER

cross-view completion (2) self-supervised learning (2) transformer architecture (2) masked image modeling (2) computer vision (1) depth estimation (1) stereo matching (1) optical flow (1) optical flow estimation (1) self-supervised pretraining (1) object pose estimation (1) 3d geometry (1) model-free approach (1) human mesh recovery (1) rgb image (1) cross-pose completion (1) stereoscopic pair (1) one-shot learning (1) 3d human understanding (1) representation learning (1)