Korean Invoice OCR Model
An OCR model designed for extracting information from Korean-language invoices. It combines deep learning with image preprocessing, targeting recognition of Korean (Hangul) characters and structured information extraction from commercial documents, achieving an F1 score of 87% on real invoice datasets.
Timeline (last 90 days)
The OCR model achieves an F1 score of 87% on a real invoice dataset.
The processing time of the OCR model is negligible.
The abstract of this paper does not disclose the specific deep learning architecture, dataset size, field-level extraction performance, or horizontal comparisons with existing mainstream OCR solutions.
All Facts (3)
The OCR model achieves an F1 score of 87% on a real invoice dataset.
50%UnverifiedThe processing time of the OCR model is negligible.
50%UnverifiedThe abstract of this paper does not disclose the specific deep learning architecture, dataset size, field-level extraction performance, or horizontal comparisons with existing mainstream OCR solutions.
50%