Summary
International Technical Conference on Circuits/Systems, Computers and Communications
2016
Session Number:P1
Session:
Number:P1-11
Optical Character Recognition Performance Comparison of Convolution Neural Network and Tesseract
Daegun Ko, Suhan Song, Kimin Kang, Seongwook Han, Juneho Yi ,
pp.871-874
Publication Date:2016/7/10
Online ISSN:2188-5079
DOI:10.34385/proc.61.P1-11
PDF download (1.3MB)
Summary:
An OCR is text recognition which is designed to extract ASCII code from an image. Our main idea is to compare two methods in OCR: one is Deep Learning based training system; another is Tesseract based pattern recognition. In this paper, we used Caffe OCR character sets to measure recognition rate and process time between Deep Learning and Tesseract. As a result, we recommend a Deep Learning method due to its performance on recognition rate. However, if process time is on priority, we recommend a Tesseract for its speed