One-Shot Template Matching for Automatic Document Data Capture

EasyChair Preprint 1763

Abstract

In this paper, we propose a novel one-shot template-matching algorithm to automatically capture data from business documents with an aim to minimize manual data entry. Given one annotated document, our algorithm can automatically extract similar data from other documents having the same format. Based on a set of engineered visual and textual features, our method is invariant to changes in position and value. Experiments on a dataset of 595 real invoices demonstrate 86.4% accuracy.

Keyphrases: Region Proposal, annotated document, automatic data capture, document analysis, document image, document processing, one-shot learning, template matching

@booklet{EasyChair:1763,
  author    = {Pranjal Dhakal and Manish Munikar and Bikram Dahal},
  title     = {One-Shot Template Matching for Automatic Document Data Capture},
  howpublished = {EasyChair Preprint 1763},
  year      = {EasyChair, 2019}}