Self-supervised Scene Text Segmentation with Object-centric Layered Representations Augmented by Text Regions.
Yibo WangYunhu YeYuanpeng MaoYanwei YuYuanping SongPublished in: ACM Multimedia (2022)
Keyphrases
- relevance feedback
- text segmentation
- scene images
- text regions
- bounding box
- multiple objects
- input image
- complex scenes
- moving objects
- information retrieval
- video sequences
- video frames
- d objects
- video images
- complex background
- optical character recognition
- object tracking
- image regions
- natural scenes
- connected components
- character recognition
- image sequences
- document images
- target object
- partial occlusion
- edge map
- object classes
- dynamic scenes
- binary images
- particle filter
- computer vision