Haotian Liu, Zhejiang University

About Me

Email: liuhaotian@zju.edu.cn
Github: haotian-liu

Hi there, I am a senior undergraduate student from Chu Kochen Honors College, Zhejiang University, China.

I am generally interested computer vision and machine learning, especially visual perception, understanding and generation, without or with little supervision. And I focus on building machines that understand the visual world the same way as humans do.

I have been working at CAD&CG State Key Lab of ZJU with Prof. Xiaogang Jin since my sophomore year. This past summer, I have spent great time at UC Davis, guided by Prof. Yong Jae Lee. Now I am working on Design2code under the supervision of Prof. Fei Wu at Institute of Artificial Intelligence of Zhejiang University with Alibaba.

Check out my CV here, and please feel free to contact me, thanks!

Education

Zhejiang University, China

Sept 2015 - June 2019

Chu Kochen Honors College, Bachelor of Engineering

GPA: 3.94/4.0 (Top 5% with Honor)
Advisor: Prof. Xiaogang Jin, Prof. Fei Wu

University of California, Davis

July - Oct 2018

ML&AI Lab, GREAT Program

Undergraduate Visiting Research Intern
Outstanding Research Performance Award

Publications

Yimeng Chen, Haotian Liu, and Lei Shi. "Operation strategy of public building: Implications from trade-off between carbon emission and occupant satisfaction." Journal of Cleaner Production 205 (2018): 629-644.
Fanyi Xiao, Haotian Liu, and Yong Jae Lee. "Identity from here, Pose from there: Self-supervised Disentanglement and Generation of Objects using Unlabeled Videos." Accepted to ICCV 2019.

Projects

Design2Code

Automatically generate layout code for a precise reproduction of massive UI designs of Alibaba product lines.

Details Comming Soon

Disentangle Model

The model learns to disentangle the identity and pose of two images, using unlabeled videos. It then generates an image composed of the identity of the first image, and the pose of the second one.

View Details

Xiao-I Service Bot

An AI Service Bot with realistic facial and skeletal animations for commercial banks.

View AI+ Service Robots

View More Projects

AR Decorator

AR Decorator based on Apple ARKit 1.0, Swift 4, and OpenGL ES 3.0. It is one of the earliest implementations of Apple ARKit with such technology stack.

It is featured with: optimized plane estimation, smooth transition, enhanced lighting, multiple gesture recognizer support, and a lot more.

Advisor: Hujun Bao

View source-code

Energy Conservation Model

The model learns from behavior statistics of students from Tsinghua University, and predicts the best energy conservation policy concerning satisfaction rate of students.

Publication: Yimeng Chen, Haotian Liu, and Lei Shi. "Operation strategy of public building: Implications from trade-off between carbon emission and occupant satisfaction." Journal of Cleaner Production 205 (2018): 629-644.