Jakarta - Humans are able to recognize a person's face with ease , but not so with computers . Well, the social networking site Facebook is currently working to develop a mechanism that promises face recognition accuracy rate " is almost equivalent to the human ability " .
Currently social networks are already able to guess who are in a photo and asking a friend to the user tagging options . However , because of its ability not maximized , users often have to manually add tags .As quoted from Mashable , Facebook API Group is developing software called DeepFace . The program is mapped in 3 -dimensional shape of the face and is able to look for specific characteristics .
The level of accuracy reached 97.25 percent called , only differs slightly than the accuracy of face recognition by humans of 97.5 percent .To develop this technology , Facebook uses visual data in the form of 4.4 million faces obtained from 4,030 users .
" System ( DeepFace ) the remaining gaps in the field of face recognition and almost reached the level of accuracy of humans , " the researchers said Facebook in a statement released by the Facebook API Group . " Performance is able to exceed the existing system and just need a little adaptation . "
If Facebook facial recognition accuracy rate can actually be made human -like capabilities , photo tagging process may later be performed fully automatically without user intervention .
No comments:
Post a Comment