Yicheng Wu

I am a Senior Applied Scientist at Amazon, working on computer vision and vision-language models for robotic automation across Amazon's fulfillment network, operating in hundreds of facilities and interacting with hundreds of millions of unique products.

Before Amazon, I was a Research Scientist at Snap Research. I received my Ph.D. from Rice University, advised by Ashok Veeraraghavan. During my Ph.D., I also interned at Google Research and Microsoft Research.

Email / Google Scholar / LinkedIn / Github

Research

	Copy or Not? Reference-Based Face Image Restoration with Fine Details Min Jin Chong, Dejia Xu, Yi Zhang, Zhangyang Wang, David Forsyth, Gurunandan Krishnan, Yicheng Wu, Jian Wang WACV, 2025 paper / code / bibtex A reference-based face restoration framework that faithfully copies fine identity-defining details (e.g., freckles, moles, scars) from a single reference image while staying semantically consistent with the degraded input.
	Energy-Efficient Adaptive 3D Sensing Brevin Tilmon, Zhanghao Sun, Sanjeev J. Koppal, Yicheng Wu, Georgios Evangelidis, Ramzi Zahreddine, Gurunandan Krishnan, Sizhuo Ma, Jian Wang CVPR, 2023 project page / paper / code / bibtex An adaptive active depth sensor that projects light only onto regions where passive stereo fails, jointly optimizing power, sensing range, and eye-safety, validated on SLM- and MEMS+DOE-based prototypes.
	Be Real in Scale: Swing for True Scale in Dual Camera Mode Rui Yu, Jian Wang, Sizhuo Ma, Sharon X. Huang, Gurunandan Krishnan, Yicheng Wu ISMAR, 2023 paper / code / bibtex Recovers true metric scale from a swinging dual-camera capture, enabling accurate real-world measurement and AR placement on mobile devices.
	Seeing Far in the Dark with Patterned Flash Zhanghao Sun, Jian Wang, Yicheng Wu, Shree Nayar ECCV, 2022 arXiv / video / code / bibtex Our proposed novel flash pattern shows significantly better imaging quality at long distances in low-light environments.
	Structured Light with Redundancy Codes Zhanghao Sun, Yu Zhang, Yicheng Wu, Dong Huo, Yiming Qian, Jian Wang arXiv, 2022 arXiv We adopt redundancy coding from the communication system to improve the robustness of the structured light system.
	3D Photo Stylization: Learning to Generate Stylized Novel Views from a Single Image Fangzhou Mu, Jian Wang, Yicheng Wu, Yin Li* (co-corresponding authors, primary mentor at Snap) CVPR, 2022 (Oral)* project page / arXiv / video / code / bibtex We address the challenging task of 3D photo stylization — generating stylized novel views from a single image given an arbitrary style.
	How to Train Neural Networks for Flare Removal Yicheng Wu, Qiurui He, Tianfan Xue, Rahul Garg, Jiawen Chen, Ashok Veeraraghavan, Jonathan T. Barron ICCV, 2021 project page / arXiv / video / code / bibtex We propose a novel semi-synthetic pipeline for data generation based on the physics model, and train neural networks to remove lens flare from a single image.
	CodedStereo: Learned Phase Masks for Large Depth-of-field Stereo Shiyu Tan, Yicheng Wu, Shoou-I Yu, Ashok Veeraraghavan (co-first author) CVPR, 2021 (Oral)* project page / arXiv / bibtex A novel end-to-end optimized stereo system with learned phase masks to achieve high-quality texture and depth reconstruction for large depth-of-field in light-limited environments.
	Deep Learning Extended Depth-of-field Microscope for Fast and Slide-free Histology Lingbo Jin, Yubo Tang, Yicheng Wu, Jackson B. Coole, Melody T. Tan, Xuan Zhao, Hawraa Badaoui, Jacob T. Robinson, Michelle D. Williams, Ann M. Gillenwater, Rebecca R. Richards-Kortum, Ashok Veeraraghavan Proceedings of the National Academy of Sciences (PNAS), 2020 paper / Rice news / bibtex AI-powered microscope could check cancer margins in minutes.
	FreeCam3D: Snapshot Structured Light 3D with Freely-Moving Cameras Yicheng Wu, Vivek Boominathan, Xuan Zhao, Jacob T. Robinson, Hiroshi Kawasaki, Aswin Sankaranarayanan, Ashok Veeraraghavan ECCV, 2020 project page / paper / video / code / bibtex We provide a structured light framework for self-localization of freely moving cameras and scene depth map estimation.
	WISHED: Wavefront Imaging Sensor with High Resolution and Depth Ranging Yicheng Wu, Fengqiang Li, Florian Willomitzer, Ashok Veeraraghavan, Oliver Cossairt (co-first author) ICCP, 2020 (Oral)* project page / paper / video / bibtex A high depth and lateral resolution 3D sensor for macroscopic rough objects by leveraging wavelength diversity and wavefront sensing.
	PhaseCam3D — Learning Phase Masks for Passive Single View Depth Estimation Yicheng Wu, Vivek Boominathan, Huaijin Chen, Aswin Sankaranarayanan, Ashok Veeraraghavan ICCP, 2019 (Oral, Best Poster Award) project page / paper / video / code / bibtex We jointly optimize the camera hardware (i.e., phase mask) and the algorithm (i.e., neural network) for depth estimation from a single image.
	WISH: Wavefront Imaging Sensor with High Resolution Yicheng Wu, Manoj Kumar Sharma, Ashok Veeraraghavan Nature - Light: Science & Applications, 2019 project page / paper / code / bibtex WISH provides wavefront reconstruction with a 10 megapixel resolution, several orders of magnitude better than commercial Shack-Hartmann sensors.
	SAVI: Synthetic Apertures for Long-Range, Sub-Diffraction Limited Visible Imaging Using Fourier Ptychography Jason Holloway, Yicheng Wu, Manoj Kumar Sharma, Oliver Cossairt, Ashok Veeraraghavan Science Advances, 2017 project page / paper / Rice news / bibtex We propose to use macroscopic Fourier ptychography as a practical means of creating a synthetic aperture for visible imaging, to achieve sub-diffraction limited resolution.
	Improvements of Measuring the Width of Fraunhofer Diffraction Fringes using Fourier Transform Yicheng Wu, Jialin Ma, Yi Yang, Ping Sun Optik-International Journal for Light and Electron Optics, 2015 paper / bibtex We propose an improved method for calculating the fringe width of Fraunhofer diffraction using Fourier transform.
	Controlling the Wave Propagation through the Medium Designed by Linear Coordinate Transformation Yicheng Wu, Chengdong He, Yuzhuo Wang, Xuan Liu, Jing Zhou European Journal of Physics, 2014 paper / bibtex Based on the principle of transformation optics, we propose to control the wave propagating by linear coordinate transformation.

Patents

Gurunandan Krishnan, Rui Yu, Yicheng Wu, Jian Wang, Sizhuo Ma. Object scale utilizing away-facing images, US20250238994A1, 2025

Shree K. Nayar, Bing Zhou, Gurunandan Krishnan, Jian Wang, Sizhuo Ma, Karl Bayer, Yicheng Wu. AR Mirror, WO2024220552A1, 2024

Numair Khalil Ullah Khan, Gurunandan Krishnan, Shree K. Nayar, Yicheng Wu. High-definition real-time view synthesis, US12112427B2, 2024

Jian Wang, Sizhuo Ma, Brevin Tilmon, Yicheng Wu, Gurunandan Krishnan, Ramzi Zahreddine, Georgios Evangelidis. Energy-efficient adaptive 3D sensing, US20240126084A1, 2024

Gurunandan Krishnan, Shree K. Nayar, Yicheng Wu. Monocular camera defocus face measuring, US20230360251A1, 2023

Yicheng Wu, Vivek Boominathan, Huaijin Chen, Aswin C. Sankaranarayanan, Ashok Veeraraghavan. Passive and single-viewpoint 3D imaging system, US20200349729A1, 2020

Yicheng Wu, Manoj Kumar Sharma, Ashok Veeraraghavan. WISH: wavefront imaging sensor with high resolution, US20200351454A1, 2020

Oliver Cossairt, Jason Holloway, Ashok Veeraraghavan, Manoj Kumar Sharma, Yicheng Wu. Synthetic apertures for long-range, sub-diffraction limited visible imaging using Fourier ptychography, US20200150266A1, 2020

Teaching

Teaching Assistant, Introduction to Computer Vision, Rice University ECE/CS, Spring 2018, 2020

Teaching Assistant, Computational Photography, Rice University ECE, Fall 2017, 2019

Website credits to Jon Barron