Betrayed by Captions: Joint Caption Grounding and Generation for Open Vocabulary Instance Segmentation.
Jianzong WuXiangtai LiHenghui DingXia LiGuangliang ChengYunhai TongChen Change LoyPublished in: CoRR (2023)
Keyphrases
- visual features
- level set
- segmentation algorithm
- image segmentation
- joint estimation
- segmentation accuracy
- segmentation method
- multiscale
- generation process
- shape prior
- bounding box
- object segmentation
- symbol recognition
- optimal segmentation
- news video
- region growing
- energy function
- medical images
- image analysis
- image processing
- fully automatic
- edge detection
- low level
- segmentation errors
- video sequences
- search engine