In financial business processes, a large number of documents need to be manually entered into the system. Document OCR can scan and identify all kinds of documents, contracts, VAT invoices, and tables in batches. In addition, Document OCR allows users to customize identification templates, greatly shortening input time and improving business processing efficiency.
A table recognition algorithm automatically matches 100+ tables (invoices, medical care, insurance documents, and so on).
A post-processing enhancement algorithm allows the recognition precision of decimal points, digits, and symbols to reach 98%.
A text separation model is used to eliminate interference such as seals, text tilt, cross-line, and superimposition. A document identification model is used to identify various document types and the front and rear sides of documents, minimizing the impacts of tilt, distortion, complex backgrounds, and imbalanced proportions.