TY - GEN
T1 - Character-independent font identification
AU - Haraguchi, Daichi
AU - Harada, Shota
AU - Iwana, Brian Kenji
AU - Shinahara, Yuto
AU - Uchida, Seiichi
N1 - Funding Information:
This work was supported by JSPS KAKENHI Grant Number JP17H06100.
PY - 2020
Y1 - 2020
N2 - There are a countless number of fonts with various shapes and styles. In addition, there are many fonts that only have subtle differences in features. Due to this, font identification is a difficult task. In this paper, we propose a method of determining if any two characters are from the same font or not. This is difficult due to the difference between fonts typically being smaller than the difference between alphabet classes. Additionally, the proposed method can be used with fonts regardless of whether they exist in the training or not. In order to accomplish this, we use a Convolutional Neural Network (CNN) trained with various font image pairs. In the experiment, the network is trained on image pairs of various fonts. We then evaluate the model on a different set of fonts that are unseen by the network. The evaluation is performed with an accuracy of 92.27%. Moreover, we analyzed the relationship between character classes and font identification accuracy.
AB - There are a countless number of fonts with various shapes and styles. In addition, there are many fonts that only have subtle differences in features. Due to this, font identification is a difficult task. In this paper, we propose a method of determining if any two characters are from the same font or not. This is difficult due to the difference between fonts typically being smaller than the difference between alphabet classes. Additionally, the proposed method can be used with fonts regardless of whether they exist in the training or not. In order to accomplish this, we use a Convolutional Neural Network (CNN) trained with various font image pairs. In the experiment, the network is trained on image pairs of various fonts. We then evaluate the model on a different set of fonts that are unseen by the network. The evaluation is performed with an accuracy of 92.27%. Moreover, we analyzed the relationship between character classes and font identification accuracy.
UR - http://www.scopus.com/inward/record.url?scp=85090094934&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85090094934&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-57058-3_35
DO - 10.1007/978-3-030-57058-3_35
M3 - Conference contribution
AN - SCOPUS:85090094934
SN - 9783030570576
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 497
EP - 511
BT - Document Analysis Systems - 14th IAPR International Workshop, DAS 2020, Proceedings
A2 - Bai, Xiang
A2 - Karatzas, Dimosthenis
A2 - Lopresti, Daniel
PB - Springer
T2 - 14th IAPR International Workshop on Document Analysis Systems, DAS 2020
Y2 - 26 July 2020 through 29 July 2020
ER -