Beyond OCR + VQA: Involving OCR into the Flow for Robust and Accurate TextVQA.
Gangyan ZengYuan ZhangYu ZhouXiaomeng YangPublished in: ACM Multimedia (2021)
Keyphrases
- optical character recognition
- character recognition
- document images
- post processing
- recognition errors
- text recognition
- preprocessing
- document processing
- error correction
- highly accurate
- computationally efficient
- image database
- text localization and recognition
- printed documents
- scanned documents
- character segmentation
- high quality
- high accuracy
- page layout
- partial occlusion
- machine vision
- natural images
- image quality
- ocr systems