PRED: Pre-training via Semantic Rendering on LiDAR Point Clouds.
Hao YangHaiyang WangDi DaiLiwei WangPublished in: CoRR (2023)
Keyphrases
- point cloud
- photorealistic
- surface reconstruction
- structure from motion
- implicit surfaces
- point cloud data
- lidar data
- point sets
- stereo camera
- laser scanner
- triangular mesh
- high quality
- multi view reconstruction
- ray tracing
- computer graphics
- feature vectors
- normal vectors
- training set
- multi view stereo
- feature selection
- training samples
- similarity measure