Arbitrary-Shape Scene Text Detection via Visual-Relational Rectification and Contour Approximation.
Chengpei XuWenjing JiaTingcheng CuiRuomei WangYuan-fang ZhangXiangjian HePublished in: IEEE Trans. Multim. (2023)
Keyphrases
- arbitrary shape
- text detection
- urban scenes
- outdoor scenes
- clustering algorithm
- high efficiency
- video analysis
- d scene
- scene understanding
- visual information
- spatial relations
- text information
- video sequences
- input image
- three dimensional
- single image
- photorealistic
- visual features
- connected components
- object recognition
- image sequences
- complex scenes
- urban environments
- computer vision
- vanishing points
- point cloud
- closed form
- camera calibration
- active contours
- viewpoint
- moving objects