報告題目：Face Hallucination via Deep Neural Networks
報告人：Dr. Xin YU，澳大利亞國立大學
摘要：Obtaining high-resolution (HR) face images plays an important role for the following face analysis tasks. In this talk, I mainly tackle the face super-resolution problem, also known as face hallucination, and propose our methods to upsample very low-resolution face images as well as recover fine details of the deteriorated faces.
These works exploit deep neural networks to super-resolve HR face images from LR counterparts in different challenging scenarios as well as to restore realistic face images from stylized portrait images. Our extensive experimental results demonstrate our proposed methods outperform the state-of-the-art.
Dr. Xin YU received his Ph.D. degree in the Department of Electronic Engineering, Tsinghua University, Beijing, China, in 2015. He also received a Ph.D. degree in the College of Engineering and Computer Science, Australian National University, Canberra, Australia, in 2018. He is currently a research fellow in Australian National University. His interests include computer vision and image processing. He has published more than 20 papers on top tier conferences and journals, such as CVPR, ICCV, ECCV, TPAMI and IJCV.