Learning 3D Representations from 2D Pre-Trained Models via Image-to-Point Masked Autoencoders.
Renrui ZhangLiuhui WangYu QiaoPeng GaoHongsheng LiPublished in: CVPR (2023)
Keyphrases
- pre trained
- image features
- learning process
- learning algorithm
- image retrieval
- image segmentation
- prior knowledge
- computer vision
- supervised learning
- image classification
- single image
- deep belief networks
- statistical model
- reinforcement learning
- keypoints
- data sets
- control signals
- pixel level
- lighting conditions
- image representation
- unsupervised learning
- feature points
- input image
- probabilistic model
- high resolution