COLA: COarse LAbel pre-training for 3D semantic segmentation of sparse LiDAR datasets.
Jules SanchezJean-Emmanuel DeschaudFrançois GoulettePublished in: CoRR (2022)
Keyphrases
- semantic segmentation
- pascal voc
- object detection
- bag of words
- weakly supervised
- superpixels
- label transfer
- conditional random fields
- object recognition
- object categories
- object class
- object classes
- training set
- sparse representation
- object classification
- test images
- image understanding
- object segmentation
- supervised learning
- training samples
- graphical models
- multi label
- co occurrence
- higher order
- high dimensional
- pairwise
- training examples