Mask TextSpotter structure
- a feature pyramid network (FPN) as backbone.
- a region proposal network (RPN) for generating text proposals.
- a Fast R-CNN for bounding boxes regression.
- a mask branch for text instance segmentation, character segmentation, and text sequence recognition.
Data flow
Training phase, RoI characterized by the RPN generated text suggestions (text proposals), and then fed to the text proposed Fast R-CNN, and generates accurate text mask branch frame candidates (candidate boxes), examples of text segmentation map (text instance segmentation maps) character segmentation map (character segmentation maps), and the sequence of text (text sequence)