ZrrSkywalker / I2P-MAE

[CVPR 2023] Learning 3D Representations from 2D Pre-trained Models via Image-to-Point Masked Autoencoders
219 stars 17 forks source link