Tencent AI supports 100 mill identity facial identity retrievals of one millisecond through YouTu Skyeye
Tencent AI supports 100 mill identity facial identity retrievals of one millisecond through YouTu Skyeye
  • By Timothy Daniel (info@koreaittimes.com)
  • 승인 2017.06.24 01:44
  • 댓글 0
이 기사를 공유합니다

Tencent YouTu Lab launched YouTu Skyeye at the 2017 Internet Good Summit on June 13(PRNewsfoto/Tencent YouTu Lab)

Tencent YouTu Lab, an AI research lab under Tencent focusing on machine learning, officially released "YouTu Skyeye", a series of products and solutions based on mass facial retrieval technology for people search, at the 2017 Internet Good Summit on June 13, further applying the technology to society to solve practical problems and enhance public welfare.

According to PRNewswire on June 23 that YouTu Skyeye supports 100 million identity facial identity retrievals, at a retrieval speed of one millisecond.

In March 2017, an anti-lost platform based on YouTu Skyeye was launched in Southeastern China's Fujian Province. On March 24, 2017, the platform successfully found a lost boy for the first time, and helped him return home. To date, the platform has successfully found 124 lost people in just about three months, and has greatly enhanced the success rate of searches for lost family members and relatives, helping the public embrace AI technology.

Such achievements are made possible thanks to YouTu Lab's increasing level of accuracy. In March 2017, YouTu Lab boosted its facial retrieval accuracy rate from 99.65% to 99.80%, which enabled it to break the world record with the facial identification database LFW. Furthermore, in April 2017, YouTu Lab topped the test using the mass facial identification database, MegaFace, with an accuracy rate of 83.29%.

It is a grim social reality that lost family members and relatives are rarely found. At the summit, YouTu Lab shared its experience on the successful finding of lost family members and relatives through the use of mass facial retrieval technology.


댓글삭제
삭제한 댓글은 다시 복구할 수 없습니다.
그래도 삭제하시겠습니까?
댓글 0
댓글쓰기
계정을 선택하시면 로그인·계정인증을 통해
댓글을 남기실 수 있습니다.

  • ABOUT
  • CONTACT US
  • SIGN UP MEMBERSHIP
  • RSS
  • 2-D 678, National Assembly-daero, 36-gil, Yeongdeungpo-gu, Seoul, Korea (Postal code: 07257)
  • URL: www.koreaittimes.com | Editorial Div: 82-2-578- 0434 / 82-10-2442-9446 | North America Dept: 070-7008-0005 | Email: info@koreaittimes.com
  • Publisher and Editor in Chief: Monica Younsoo Chung | Chief Editorial Writer: Hyoung Joong Kim | Editor: Yeon Jin Jung
  • Juvenile Protection Manager: Choul Woong Yeon
  • Masthead: Korea IT Times. Copyright(C) Korea IT Times, All rights reserved.
ND소프트