Sizhuo Ma

Sizhuo Ma

Research Scientist

Snap Research

Biography

I am a Research Scientist at Snap Research. My research interests lie in computer vision and computational imaging. I received my PhD in Computer Sciences at UW-Madison, advised by Professor Mohit Gupta.

My PhD work focuses on solving geometry, motion-related computer vision problems using novel computational cameras. For example, how can we accurately recover minuscule motion of objects? How can we take a clear, sharp image of a extremely dark and moving scene? I develop novel solutions to these problems using light field cameras, structured light and single-photon cameras.

Download my CV.

Interests
  • Computer Vision
  • Computational Photography
  • Computational Imaging
Education
  • PhD in Computer Sciences, 2022

    University of Wisconsin-Madison

  • MS in Computer Sciences, 2016

    University of Wisconsin-Madison

  • BS in Computer Science, 2014

    Shanghai Jiao Tong University

News

June 2024: I will attend CVPR 2024 in Seattle, WA.

May 2023: One paper accepted to SIGGRAPH 2023!

April 2023: One paper accepted to MobiCom 2023!

Februrary 2023: One paper accepted to CVPR 2023!

October 2022: One paper accepted to WACV 2023!

June 2022: I will attend CVPR 2022 in New Orleans, LA.

May 2022: I received 2022 Outstanding Graduate-Student Research Award! Thanks for everyone who has collaborated with me or supported my research.

February 2022: I joined Snap Research as a Research Scientist!

December 2021: I passed my PhD oral defense! Thanks everyone for their support!

December 2020: I received 2020 Snap Research Fellowship!

June 2020: Quanta Burst Photography was reported by UW-Madison News and EPFL News.

May 2020: Our paper Quanta Burst Photography was featured in SIGGRAPH Technical Papers highlights.

Projects

RobustSAM: Segment Anything Robustly on Degraded Images

CVPR 2024
Make Segment Anything model robust to image degradations
RobustSAM: Segment Anything Robustly on Degraded Images
DisCO: Portrait Distortion Correction with Perspective-Aware 3D GANs

IJCV, 2024
Correct perspective disotrtion of portrait images with 3D GAN
DisCO: Portrait Distortion Correction with Perspective-Aware 3D GANs
Personalized Restoration via Dual-Pivot Tuning

arXiv, 2023
Improving face restoration with reference images for better identity
Personalized Restoration via Dual-Pivot Tuning
Be Real in Scale: Swing for True Scale in Dual Camera Mode

ISMAR 2023
Estimate the metric scale of user faces by simply swinging the phone, utilizing front and rear cameras simultaneously
Be Real in Scale: Swing for True Scale in Dual Camera Mode
Seeing Photons in Color

SIGGRAPH 2023
Color filter and algorithm design for single-photon color imaging in low light
Seeing Photons in Color
QfaR: Location-Guided Scanning of Visual Codes from Long Distances

MobiCom 2023
A novel location-guided approach that extends the scanning distance of QR codes by 4x or more
QfaR: Location-Guided Scanning of Visual Codes from Long Distances
Energy-Efficient Adaptive 3D Sensing

CVPR 2023
Energy-efficient and eye-safe active 3D sensing that is adapted to the scene and application
Energy-Efficient Adaptive 3D Sensing
Privacy-Preserving Visual Localization with Event Cameras

arXiv, 2022
Privacy-Preserving Visual Localization with Event Cameras
Burst Vision Using Single-Photon Cameras

WACV 2023
Exploring the capabilities of SPAD sensors for a wide gamut of real-world computer vision tasks including object detection, pose estimation, SLAM, text recognition and so on
Burst Vision Using Single-Photon Cameras
Single-Photon Structured Light

CVPR 2022
Structured light 3D imaging enabled at extreme speeds and challenging scenarios using single-photon cameras and digital micro-mirror devices.
Single-Photon Structured Light
Inertial Safety from Structured Light

ECCV 2020
A novel scene representation that enables fast detection of obstacles in scenarios involving camera or scene motion using single-shot structured light
Inertial Safety from Structured Light
Quanta Burst Photography

SIGGRAPH 2020
A computational imaging technique with single-photon cameras enables ultra-low light photography
Quanta Burst Photography
3D Scene Flow from 4D Light Field Gradients

ECCV 2018 oral presentation, selected for IJCV Special Issue on Best of ECCV
Recover high-precision dense scene flow from light fields
3D Scene Flow from 4D Light Field Gradients

Publications

(2024). DSL-FIQA: Assessing Facial Image Quality via Dual-Set Degradation Learning and Landmark-Guided Transformer. CVPR 2024.

PDF Project Video

(2024). RobustSAM: Segment Anything Robustly on Degraded Images. CVPR 2024.

PDF Code Project Video

(2024). DisCO: Portrait Distortion Correction with Perspective-Aware 3D GANs. IJCV, 2024.

PDF Project

(2024). Sagiri: Low Dynamic Range Image Enhancement with Generative Diffusion Prior. arXiv, 2024.

PDF Code Project

(2023). Clearer Frames, Anytime: Resolving Velocity Ambiguity in Video Frame Interpolation. arXiv, 2023.

PDF Code Project