The Face Group has been working on face analysis since 1999, and has published more than 300 papers on spoofing detection, face detection, face landmark detection, face attributes prediction, facial attribute editing, and face recognition. These works have been cited for more than 25000 times (Google Scholar). Some of these works won the 2nd class of China’s National Science and Technology Progress Award in 2005, and 2nd class of China’s State Natural Science Award in 2015. Moreover, a few face recognition and analysis works have been successfully commercialized into China’s Ministry of Public Security Exit-Entry Administration, Provincial Public Security Bureau, Labour and Social Security Bureau, and the cellphone. In August 2016, Beijing Seetatech Technology Co., Ltd (Seetatech) is established especially for commercialization of face recognition techniques.
The face group mainly focuses on the following two aspects:
1) Theory and Methodology: mainly study the fundamental problems of computer vision and pattern recognition, including discriminative feature learning, deep learning with defect data (e.g. small number of data, weakly supervised data, nosily labeled data), self-supervised visual learning, generative learning and adversarial attack, transfer learning.
2) Application and Methods: mainly study the fundamental problems of face application, including face detection and tracking, facial landmark detection, face recognition, 3D face reconstruction, facial expression analysis, lip reading, facial attributes estimation.