Simpler is Better: Few-shot Semantic Segmentation with Classifier Weight Transformer.
Zhihe LuSen HeXiatian ZhuLi ZhangYi-Zhe SongTao XiangPublished in: CoRR (2021)
Keyphrases
- semantic segmentation
- street scenes
- conditional random fields
- label transfer
- superpixels
- training data
- scene classification
- object categories
- weakly supervised
- object class
- video sequences
- object classes
- feature space
- training examples
- visual features
- feature selection
- support vector machine
- machine learning
- learning algorithm
- decision trees
- pascal voc
- image set
- small number
- natural scenes
- information extraction
- key frames
- probabilistic model
- training set
- object detection
- object recognition
- co occurrence
- segmentation algorithm
- natural images