Presentation | 2016-03-11 A Feasibility Study of Scene Text Recognition Based on Filtered Channel Features Rui Kasuya, Takuro Oki, Ryusuke Miyamoto, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | Scene Text Recognition is one of the most challenging task in the field of the image recognition, recently. Several kinds of window-based schemes have tried to improve the recognition accuracy but it is difficult to improvethe recognition accuracy without increasing processing cost when a multi-class classifier is used for this task. Con-sidering this problem, the authors have proposed a novel approach that consists of two kinds of opearations: textlocalization and text recognition. In the proposed approach, text regions are detected by a two-class classifier, and then, a multi-class classifier is applied to determined what characters are included in the detected regions. However, a multi-class classifier requried for our approach has not been constructed yet. Therefore, this paper aims to construct a multi-class classifier suitable for the proposed approach by applying the Filtered Channel Features for the training process of a multi-class classifier. Experimental results using the ICDAR2013 dataset and synthesized images created by the authors show that the precision becomes high but the recall are not improved drastically. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | scene text recognition / Filtered Channel Features / multi-class classification |
Paper # | SIS2015-65 |
Date of Issue | 2016-03-03 (SIS) |
Conference Information | |
Committee | SIS |
---|---|
Conference Date | 2016/3/10(2days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | Tokyo City Univ. |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | Soft computing, etc. |
Chair | Mitsuji Muneyasu(Kansai Univ.) |
Vice Chair | Hirokazu Tanaka(Hiroshima City Univ.) / Takayuki Nakachi(NTT) |
Secretary | Hirokazu Tanaka(Nagoya City Univ.) / Takayuki Nakachi(Toshiba) |
Assistant | Hiroyuki Tsuji(Kanagawa Inst. of Tech.) / Hakaru Tamukoh(Kyushu Inst. of Tech.) |
Paper Information | |
Registration To | Technical Committee on Smart Info-Media System |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | A Feasibility Study of Scene Text Recognition Based on Filtered Channel Features |
Sub Title (in English) | |
Keyword(1) | scene text recognition |
Keyword(2) | Filtered Channel Features |
Keyword(3) | multi-class classification |
1st Author's Name | Rui Kasuya |
1st Author's Affiliation | Meiji University(Meiji Univ.) |
2nd Author's Name | Takuro Oki |
2nd Author's Affiliation | Meiji University(Meiji Univ.) |
3rd Author's Name | Ryusuke Miyamoto |
3rd Author's Affiliation | Meiji University(Meiji Univ.) |
Date | 2016-03-11 |
Paper # | SIS2015-65 |
Volume (vol) | vol.115 |
Number (no) | SIS-505 |
Page | pp.pp.99-104(SIS), |
#Pages | 6 |
Date of Issue | 2016-03-03 (SIS) |