On 31st March 2021, the Institute of Statistics and Computational Intelligence of DST invited the AI scientist Dr. Wu Zhili to deliver a lecture on Face AI. In the talk, he analyzed the great opportunities and challenges of face AI.

Dr. Wu Zhili

Dr. Wu Zhili is an AI scientist and the founder of Shenzhen Yiyuan Intelligence Technology Co. Ltd. He studied at Tsinghua University and Hong Kong Baptist University. After obtaining a doctorate degree from Hong Kong Baptist University, he then completed a postdoctoral degree at the University of Leeds in the United Kingdom. He founded Yiyuan Intelligence Technology in 2017, which took medical-grade imaging AI as the entry point and formed a series of AI models based on skin big data. The company has served almost 2,000 customers in the industry through API platforms, SDKs and SaaS software.

According to Dr. Wu, face AI not only refers to face recognition technology, but also includes skin diagnosis, feature measurement, virtual metahuman, makeup test, face changing, 3D face technology, beauty selfie, and so on. He first talked about the economic value of face AI. Skin diagnosis and features measurement have great industrial value among young customers in the beauty industry. Face recognition, the most commonly used face AI, has huge market assets related to the security industry. Both virtual metahuman and makeup tests can be frequently used in marketing channels. Face changing and beauty selfie applications earn a lot through entertaining apps. At the same time, Dr. Wu mentioned various challenges facing AI, such as privacy leakage, data ownership dispute, low test accuracy, and so on.

Dr. Wu receiving a souvenir from Prof. Shi Hongjian

Dr. Wu then introduced AI Face Feature Analysis. Vision AI and Recommender System are two important technologies. Vision AI technology consists of four parts—doing face detection and alignment, selecting key points on human face as training set, classifying the feature, applying statistical and machine learning methods to solve specific tasks. After elaborating on key technology, Dr. Wu introduced the current status of AI skin testing industry and situations of his company.

Group photo with audiences

At the end of the lecture, teachers and students raised some questions about the upper limit of the modifiable ratio of facial photo, the method of confirming ID information, the way to identify skin age through face photo, and the method of quantifying beauty, etc. Dr. Wu answered patiently one by one.

Faculty and student asking questions

Dr. Wu had a meeting with juniors in statistics

Reporters: Daisy Huang & Lily Peng

Photographers: Lily Peng

Editors: Garbo Hu & Quincy Zhang

(From STAT Programme)