Haozheng Yu

Hi, thanks for visiting my website! I am a PhD student at Cornell University, advised by Professor Jennifer J. Sun. I am interested in building AI systems that can understand, reconstruct and synthesize the dynamic 3D world around us.

Previously, I was a Machine Learning Engineer at Irreverent Labs. I spent a wonderful year at the Media Lab, Tencent America. I completed my M.S. in Computer Science at the University of Minnesota - Twin Cities, where I was fortunate to be advised by Prof. Hyun Soo Park. Prior to that, I received my Bachelor's degree in Computer Science and Technology from Central China Normal University.

Email  /  Google Scholar  /  LinkedIn  /  Github

Inspired by Prof. Kyunghyun Cho and Prof. Wei-Chiu Ma, I commit 30 minutes to 1 hour per week to meet with undergraduates, early-career graduate students, or whoever is in need about research, career development, potential research opportunities, or graduate student life. Please feel free to reach out via email if you would like to chat :)

We introduce the panel representation of panoramas to solve the indoor 360 understanding tasks.

We present a probabilistic epipolar constraint to learn a dense keypoint detector from unlabeled multiview images.

I designed a bootstrapping strategy to fine-tune a Mask-RCNN on monkey data with only sparse keypoint annotations. I implemented a voxel-based visual hull reconstruction to generate 3D monkey mesh from multiview segmentation results.

CS 4782: Intro to Deep Learning (Spring 2025), Cornell University - Teaching Assistant

Thanks Jon Barron for this great template!