FreeSeg: Free Mask from Interpretable Contrastive Language-Image Pretraining for Semantic Segmentation.
Yi LiHuifeng YaoHualiang WangXiaomeng LiPublished in: CoRR (2022)
Keyphrases
- semantic segmentation
- input image
- image data
- multiscale
- bounding box
- image features
- superpixels
- weakly supervised
- scene classification
- image segmentation
- image content
- conditional random fields
- image representation
- object categories
- region of interest
- single image
- pixel level
- image regions
- image retrieval
- image classification
- natural language
- image matching
- long range
- object classes
- image set
- keypoints
- segmentation method
- object recognition
- high resolution