Zexin HE

I am currently a first-year Ph.D. student at the Chinese University of Hong Kong, advised by Prof. Jiaya Jia.

I received my B.Eng. degree at the Honors College of Beihang University, where I was a research intern at Cola Laboratory. I've also spent a wonderful time as an intern at SenseTime Research.

My research interests lie primarily in Computer Vision and Machine Learning, mainly focusing on Radiance Fields and Generative Models. I also have a broad interest in Bayesian Machine Learning and Computational Arts.

Email  /  Google Scholar  /  Github

We present a controllable scene stylization paradigm to stylize a 3D scene with a single stylized 2D view as reference. Ref-NPR propagates 2D style to 3D radiance fields and obtains high visual quality and visual correspondence.

We build a mega-scale information-dense dataset for both classification and detection with active learning, which is 4 times larger than ImageNet and 2 times larger than Object365.

Generate background music from an arbitrary video clip by establishing the correspondence between music features and video features as well as proposing a targeted music symbolic representation.

Last updated: Dec 2022
Web page design credit to Jon Barron
Avatar style credit to Shaoteng Liu