China’s iQIYI is sharing its facial recognition research at the IEEE’s International Conference on Computer Vision (ICCV). The company has published a paper that specifically addresses unlabeled facial recognition data, and details a method that does not require any manual labeling and will allow that data to be processed more efficiently.
Doing so will help make facial recognition technology more accurate, especially when dealing with variables like pose, makeup, and lighting that can be difficult to measure in a natural, uncontrolled environment. As a provider of online entertainment services, iQIYI is hoping to use the technology to develop more engaging content for consumers.
“Multimodal person identification technology is a top priority for iQIYI researchers,” said iQIYI Senior Director Liu Junhui. “Going forward, iQIYI will continue exploring the endless possibilities of entertainment experience by breaking through technological boundaries.”
iQIYI previously hosted the Celebrity Video Identification Challenge. The company was still stargazing at ICCV, and released a new VID-FACE data set that includes more than 6.3 million photos and hundreds of thousands of video clips of more than 10,000 different celebrities.
IEEE, meanwhile, recently published a new Generation AI study that found growing support for the use of AI technology in children’s healthcare. The organization regularly solicits papers on a range of different topics.
ICCV 2019 runs from October 27 to November 2 in Seoul, South Korea.
October 31, 2019 – by Eric Weiss