method: Conv_Attention2019-06-02

Authors: Daehyun Nam

Description: Convolution-based feature extractor followed by patch-based prediction with attention. Feature etractor is pre-trained with ImageNet. Roll-back method was applied to enhance low-level features. CE loss with label smoothing.

Confusion Matrix

Detection
ArabicLatinChineseJapaneseKoreanBanglaHindiSymbolsNone
GTArabic47831562049303738290
Latin32453107865284693043561415160
Chinese4107411946330116100
Japanese29531356691413838351160
Korean23699188403114736275690
Bangla243512524472740
Hindi5655144864024210
Symbols181721941872837220
None000000000