PReFormer: A memory-efficient transformer for point cloud semantic segmentation.
Perpetual Hope AkwensiRuisheng WangBo GuoPublished in: Int. J. Appl. Earth Obs. Geoinformation (2024)
Keyphrases
- point cloud
- memory efficient
- semantic segmentation
- conditional random fields
- superpixels
- structure from motion
- weakly supervised
- scene classification
- object categories
- point sets
- pascal voc
- object class
- object classes
- sparse reconstruction
- object segmentation
- image set
- bag of words
- object detection
- multiscale
- image classification
- image features
- face recognition