Fetching the paper…

Learning from 2D: Contrastive Pixel-to-Point Knowledge Transfer for 3D Pretraining · Around