ContourNet: Taking a Further Step toward Accurate Arbitrary-shaped Scene Text Detection.
Yuxin WangHongtao XieZhengjun ZhaMengting XingZilong FuYongdong ZhangPublished in: CoRR (2020)
Keyphrases
- text detection
- arbitrary shaped
- urban scenes
- natural scene images
- outdoor scenes
- single image
- d scene
- text information
- video analysis
- input image
- high accuracy
- video sequences
- complex scenes
- object recognition
- highly accurate
- point cloud
- scene understanding
- urban environments
- packing problem
- connected components
- multi band
- geometric constraints
- vanishing points
- arbitrary shape
- complex background
- structure from motion
- vision system
- moving objects
- image sequences