ZegCLIP: Towards Adapting CLIP for Zero-shot Semantic Segmentation.
Ziqin ZhouYinjie LeiBowen ZhangLingqiao LiuYifan LiuPublished in: CVPR (2023)
Keyphrases
- semantic segmentation
- object categories
- street scenes
- conditional random fields
- superpixels
- label transfer
- object classes
- object recognition
- scene classification
- object detection
- pascal voc
- object class
- weakly supervised
- object classification
- training examples
- generative model
- visual words
- image understanding
- image set
- feature extraction
- image features
- low level features
- pixel level
- object segmentation
- segmentation algorithm
- supervised learning
- support vector
- multiscale