Yichao Zhou

Ph.D. in Computer Science

University of California, Berkeley


I joined Apple after graduation. Previously, I received my Ph.D. degree in Computer Science at UC Berkeley. My dissertation is on “Learning to Detect Geometric Structures from Images for 3D Parsing”.

I am interested in 3D vision, computer graphics, and artificial intelligence. My recent research focuses on integrating geometric concepts and data-driven approaches into a unified framework. To this end, I design methods to extract high-level geometric structures from images, such as lines, wireframes, vanishing points, and symmetries, and use them to parse scenes into high-quality compact CAD-like 3D models and achieve robust localization.


  • 3D Vision
  • Computer Graphics
  • Artificial Intelligence


  • Ph.D. in Computer Science, 2020

    University of California, Berkeley

  • B.Eng in Computer Science and Technology, 2015

    Tsinghua University

Blog Posts

I will discuss the generalized approach to sample points on an arbitrary geometry.
I will discuss the math and intuition behind the 3D altitude filter, linear Kalman filter, extended Kalman filter, error-state Kalman filter, and Lie group calculus for deriving the math for orientation estimation.
I will summarize my superficial understand of the risk in paper gold.
I will summarize what vanishing points (and their “Gaussian sphere representation”) are, how to represent them, what information they encode, how to find them, and why they are useful.


”*” indicates equal contribution

(2022). Rapid Face Asset Acquisition with Recurrent Feature Alignment. SIGGRAPH Asia 2022.


(2021). VaPiD: A Rapid Vanishing Point Detector via Learned Optimizers. CVPR 2021.


(2021). NeRD: Neural 3D Reflection Symmetry Detector. CVPR 2021.

Preprint Code

(2020). HoliCity: A City-Scale Data Platform for Learning Holistic 3D Structures. Technical Report. arXiv:2008.03286 [cs.CV].

Preprint Code Project

(2020). Learning to Detect 3D Reflection Symmetry for Single-View Reconstruction. Technical Report. arXiv:2006.10042 [cs.CV].

Preprint Code

(2020). ManifoldPlus: A Robust and Scalable Watertight Manifold Surface Generation Method for Triangle Soups. Technical Report. arXiv:2005.11621 [cs.GR].

Preprint Code

(2019). NeurVPS: Neural Vanishing Point Scanning via Conic Convolution. NeurIPS 2019.

Preprint Code Poster Slides

(2019). End-to-End Wireframe Parsing. ICCV 2019.

Preprint Code Poster

(2019). FrameNet: Learning Local Canonical Frames of 3D Surfaces from a Single RGB Image. ICCV 2019.

Preprint Code

(2018). Learning to Reconstruct 3D Manhattan Wireframes from a Single Image. ICCV 2019 (Oral Presentation).

Preprint Code Dataset Poster Video

(2018). QuadriFlow: A Scalable and Robust Method for Quadrangulation. SGP 2018 (Best Paper), Computer Graphics Forum.

PDF Code Dataset Demo Online Tool

(2016). Parallel Computational Protein Design. Methods in Molecular Biology.

PDF Code

(2015). Massively Parallel A* Search on a GPU. AAAI 2015 (Oral Presentation).

PDF Code Slides Appendix

(2014). An Efficient Parallel Algorithm for Accelerating Computational Protein Design. ISMB 2014, Bioinformatics.

PDF Code Dataset Poster Slides



February 2021 – Present
California, USA

Research Scientist

Apple Inc.

I currently work in the SPG org. Preivously, I was a 3D Vision Engineer in the SLAM team.
May 2019 – August 2019
3000 El Camino Real, Palo Alto, CA 94306

Research Intern

Bytedance Inc.

I created a real-world data platform of urban environments by aligning the city CAD model to the street-view panorama images. I was advised by Yi Ma and Linjie Luo.
May 2018 – August 2018
345 Park Ave, San Jose, CA 95110

Research Intern

Adobe Research

I did the research on how to reconstruct a 3D scene for urban environments by utilizing their global regularities and symmetries. I enjoyed my time with my advisors Qi Sun, Zhili Chen, Li-Yi Wei, and Kalyan Sunkavalli.
May 2017 – August 2017
523 Ocean Front Walk, Venice, CA 90291

Research Intern

Snapchat Inc.

I worked on a project about algorithm-based Beatles-style music composition with the help from an outstanding composer and percussionist Sam Young at UCLA. I was advised by Wei Chu and Xin Chen.
May 2016 – August 2016
1901 S Bascom Ave #1300, Campbell, CA 95008

Research Intern

Bellus 3D Inc.

During the internship, I implemented a software to deform a face mesh to fit the point cloud of a scanned face along with the Morphable Face Model. I was advised by James Davis and Jing Liu.
September 2015 – December 2020
University of California, Berkeley, CA 94720

Research Assistant

University of California, Berkeley

I received my Ph.D. degree here. My advisor is Prof. Yi Ma. My dissertation is on “Learning to Detect Geometric Structures from Images for 3D Parsing”.
December 2013 – June 2015
30 Shuangqing Rd, Haidian District, Beijing 100084

Research Assistant

Tsinghua University

Advised by Jianyang Zeng, I did research related to the proteins’ 3D structures in the Machine Learning and Computational Biology Group at IIIS when I was an undergraduate. I designed fast and massively parallel combinotorial search algorithms and tested them on the protein design problems. I also implemented a CUDA version of RELION to accelerate the 3D reconstruction of protein structures from cryo-microscopy images, which later turned into my undergraduate thesis.
August 2013 – November 2013
5 Danling St, Haidian District, Beijing 100080

Research Intern

Microsoft Research Asia

I was a research intern in the Natural Language Processing Group at MSRA, advised by Jianwen Zhang. I studied method of applying Markov Logic Network for the knowledge database completion problem to fill in the missing entires in Freebase.


I was a teaching assistant of the following courses at UC Berkeley:



Deep Learning



Blender 3D