WordSup: Exploiting Word Annotations for Character based Text Detection.
Han HuChengquan ZhangYuxuan LuoYuzhuo WangJunyu HanErrui DingPublished in: CoRR (2017)
Keyphrases
- text detection
- scene text
- text lines
- natural scene images
- object recognition
- document images
- connected components
- co occurrence
- video analysis
- text information
- metadata
- optical character recognition
- keywords
- scene images
- urban scenes
- image annotation
- multi band
- text regions
- outdoor scenes
- complex background
- structure from motion