• Àüü
  • ÀüÀÚ/Àü±â
  • Åë½Å
  • ÄÄÇ»ÅÍ
´Ý±â

»çÀÌÆ®¸Ê

Loading..

Please wait....

±¹³» ³í¹®Áö

Ȩ Ȩ > ¿¬±¸¹®Çå > ±¹³» ³í¹®Áö > Çѱ¹Á¤º¸°úÇÐȸ ³í¹®Áö > Á¤º¸°úÇÐȸ³í¹®Áö (Journal of KIISE)

Á¤º¸°úÇÐȸ³í¹®Áö (Journal of KIISE)

Current Result Document :

ÇѱÛÁ¦¸ñ(Korean Title) RDID-GAN: ºñ½Äº°È­ À̹ÌÁö µ¥ÀÌÅÍ º¹¿øÀ» ÅëÇÑ È¿°úÀûÀÎ ÇнÀµ¥ÀÌÅÍ »ý¼º
¿µ¹®Á¦¸ñ(English Title) RDID-GAN: Reconstructing a De-identified Image Dataset to Generate Effective Learning Data
ÀúÀÚ(Author) ¿À¿ø¼®   ¹è°­¹Î   ¹èÀ¯¼®   Wonseok Oh   Kangmin Bae   Yuseok Bae  
¿ø¹®¼ö·Ïó(Citation) VOL 48 NO. 12 PP. 1329 ~ 1334 (2021. 12)
Çѱ۳»¿ë
(Korean Abstract)
ÃÖ±Ù ¿©·¯ »çȸ ¹®Á¦µéÀ» ¿¹¹æ ¹× ½Å¼ÓÇÏ°Ô ´ëóÇϱâ À§ÇØ CCTV°¡ ¼³Ä¡µÇ°í ÀÖ°í ÀΰøÁö´ÉÀ» È°¿ëÇØ À̸¦ È¿°úÀûÀ¸·Î ó¸®ÇÏ´Â ¹æ¾ÈÀÌ ¿¬±¸µÇ°í ÀÖ´Ù. ÇÏÁö¸¸, CCTV¿¡¼­ ¼öÁýÇÑ µ¥ÀÌÅÍ´Â °³ÀÎÁ¤º¸ ħÇØÀÇ ¿ì·Á°¡ ÀÖ¾î ºñ½Äº°È­ ÀÛ¾÷ ¾øÀÌ´Â ÀÚÀ¯·Ó°Ô »çȸ¹®Á¦ ÇØ°áÀ» À§ÇÑ ¸ðµ¨À» ¿¬±¸Çϴµ¥ »ç¿ëÇÒ ¼ö ¾ø´Ù. µû¶ó¼­, º» ³í¹®¿¡¼­´Â RDID-GANÀ» Á¦¾ÈÇÏ¿© ºñ½Äº°È­µÈ »ç¶÷ÀÇ ¾ó±¼À» ÀÓÀÇ·Î º¹¿øÇÏ¿© °³ÀÎ Á¤º¸ ħÇØÀÇ ¿ì·Á¸¦ ÁÙÀÌ°í ³×Æ®¿öÅ© ÇнÀ¿¡µµ ºÎÁ¤ÀûÀÎ ¿µÇâÀ» ÁÖÁö ¾Ê´Â È¿°úÀûÀÎ µ¥ÀÌÅͼ Á¦ÀÛ ¹æ¾ÈÀ» Á¦¾ÈÇÑ´Ù. RDID-GANÀº attention moduleÀ» È°¿ëÇØ ºñ½Äº°È­µÈ ºÎºÐ¿¡ ÁýÁßÇÏ¿© ÇÕ´çÇÑ °á°ú¸¦ »ý¼ºÇÒ ¼ö ÀÖµµ·Ï ÇÏ¿´´Ù. ¿ì¸®´Â ½ÇÇèÀ» ÅëÇØ ÇØ´ç ¸ðµ¨°ú ±âÁ¸ÀÇ Á¦¾ÈµÈ image-to-image º¯È¯ ¸ðµ¨À» Á¤¼ºÀû ¹× Á¤·®ÀûÀ¸·Î ºñ±³ÇÏ¿´´Ù.
¿µ¹®³»¿ë
(English Abstract)
Recently, CCTVs have been installed to prevent or handle various social problems, and there are many efforts to develop visual surveillance systems based on deep neural networks. However, the datasets collected from CCTVs are inappropriate to train models due to privacy issues. Therefore, in this paper, we proposed RDID-GAN, an effective dataset de-identification method that can remove privacy issues and negative effects raised by modifying the dataset using a de-identification procedure. RDID-GAN focuses on a de-identified region to produce competitive results by adopting the attention module. Through the experiments, we compared RDID-GAN and the conventional image-to-image translation models qualitatively and quantitatively.
Å°¿öµå(Keyword) Àû´ëÀû »ý¼º¸Á   ºñ½Äº°È­ µ¥ÀÌÅͼ   »ç¶÷ Àνı⠠ ½Ã°¢ °¨½Ã   image-to-image º¯È¯   GAN   de-identified dataset   person detector   visual surveillance   image-to-image translation  
ÆÄÀÏ÷ºÎ PDF ´Ù¿î·Îµå