Reducing Annotation Effort by Identifying and Labeling Contextually Diverse Classes for Semantic Segmentation Under Domain Shift.
Sharat AgarwalSaket AnandChetan AroraPublished in: WACV (2023)
Keyphrases
- semantic segmentation
- street scenes
- object classes
- annotation effort
- active learning
- conditional random fields
- superpixels
- object categories
- pascal voc
- scene classification
- label transfer
- weakly supervised
- object detection
- object class
- image understanding
- hidden markov models
- image set
- bag of words
- domain specific
- image features
- pairwise
- object recognition
- long range
- feature extraction
- input image
- training data