Current Result Document :
ÇѱÛÁ¦¸ñ(Korean Title) |
°ø°£ ºñ´ëĪ ¾îÅÙ¼Ç ±â¹ýÀ» È°¿ëÇÑ ½Ç¼¼°è ½ºÇªÇÎ ¹æÁö |
¿µ¹®Á¦¸ñ(English Title) |
Learning real-world anti-spoofing leveraging spatialasymmetric attention |
ÀúÀÚ(Author) |
S M A Sharif
±è¼ºÁØ
¹Ú»ó¹Î
S M A Sharif
SungJun Kim
SangMin Park
|
¿ø¹®¼ö·Ïó(Citation) |
VOL 49 NO. 02 PP. 0556 ~ 0558 (2022. 12) |
Çѱ۳»¿ë (Korean Abstract) |
|
¿µ¹®³»¿ë (English Abstract) |
Face anti-spoofing (FAS) consider one of the most crucial attributes of a secure face recognition system. This study proposes a novel FAS method to handle real-world spoofing attacks. Our method comprises a deep network with spatial-asymmetric attention to refine extracted features of a VGG-16-based backbone to outperform existing methods. Also, we collected a real-world RGB spoof dataset to learn and evaluate real-world spoofing attacks. The experiment analysis demonstrates that the proposed method can substantially enhance spoof detection in real-world scenarios without incorporating specialized hardware. |
Å°¿öµå(Keyword) |
|
ÆÄÀÏ÷ºÎ |
PDF ´Ù¿î·Îµå
|