CyberLink and Perfect Corp. recently announced the winners of the second annual “AI Meets Beauty Challenge.” The two organizations hosted the international competition in collaboration with Taiwan’s National Chiao Tung University.
The “AI Meets Beauty Challenge” was established in an effort to develop smart retail solutions that will improve the customer experience in the beauty industry. This year’s challenge asked competitors to identify every single product in a Perfect-500K dataset that included more than 500,000 images of various beauty products.
The winners were revealed at the ACM Multimedia Conference in Nice, France, where the top prize went to the team from the University of Science and Technology of China. The team beat out 54 other entrants with an AI image recognition solution that was able to reduce background noise and place the focus on the product to make it easier to identify.
“We are excited to see so many talented AI researchers striving to achieve technology breakthrough,” said Perfect Corp. Founder and CEO Alice Chang. “We believe research and advanced technology developments like those submitted in this contest will empower the beauty industry’s digital transformation and reshape the customer’s journey.”
“In addition to facial recognition, AI subject identification is an important technology for smart retail and a key focus for our company,” said CyberLink CEO Jau Huang. “This competition contributes to CyberLink’s search for partners to develop future smart retail solutions.”
CyberLink is the developer of the FaceMe facial recognition engine, and has regularly highlighted the retail applications of its platform. Earlier this year, FaceMe appeared in a customer service robot from the Syscom Group, and it has also been integrated into a depth sensing chip from Altek. The algorithm posted a top tier result in a recent round of NIST testing.
The “AI Meets Beauty Challenge” runners-up were the teams from the Guangdong University of Technology and the Guangzhou Shiyuan Electronic Technology Company.
Follow Us