About Me

I am a final-year PhD student at the Department of Computer Science and Technology, University of Cambridge, supervised by Prof. Srinivasan Keshav FRS and Prof. Andrew Blake FRS. I am affiliated with the UKRI Centre for Doctoral Training in AI for Environmental Risks (AI4ER).

My research sits at the intersection of computer vision and Earth observation, where I develop synthetic data and physics-informed machine learning methods for Earth intelligence. It builds on a simple but powerful idea: the physical models that describe how satellite imagery is formed — radiative transfer — are essentially graphics renderers. This connection between physics-based simulation and computer vision opens a new frontier for AI in science.

Beyond research, I enjoy choral singing and Chinese calligraphy. If you are interested in my work or would like to connect, please feel free to reach out!

Research

Three threads of my PhD are converging into a single pipeline — from virtual forest scenes, to synthetic satellite imagery, to interpretable evaluation of AI models:

  • Synthetic data for 3D forest vision. CAMP3D uses game-engine forest scenes and virtual drone surveys to generate labelled laser scanning data at a scale and diversity exceeding any real-world forest dataset, for training 3D segmentation algorithms for forest carbon quantification.

  • Physics-informed representation learning. PILA integrates radiative transfer models into neural networks to retrieve interpretable biophysical variables from satellite imagery, enabling scalable and trustworthy environmental monitoring.

  • Probing foundation models with simulation. Radiative transfer is a physically-grounded renderer: by extending CAMP3D to render multi-spectral satellite imagery with full ground truth, and building on PILA’s framework, I am designing probes that test what geo-foundation models actually learn.

My ambition is to build an open-source toolkit that makes synthetic data for Earth observation as accessible and transformative as it has been for self-driving.

Publications

PILA: Physics-Informed Low Rank Augmentation for Interpretable Earth Observation
Yihang She, Andrew Blake, Clement Atzberger, Srinivasan Keshav
Preprint, under review at the Journal of Machine Learning Research
Scaling Up Forest Vision with Synthetic Data
Yihang She, Andrew Blake, David Coomes, Srinivasan Keshav
International Journal of Computer Vision, 2026 (accepted)
SPREAD: A Large-Scale, High-Fidelity Synthetic Dataset for Multiple Forest Vision Tasks
Zhengpeng Feng, Yihang She, Srinivasan Keshav
Ecological Informatics, 2025
Fast Hierarchical Learning for Few-Shot Object Detection
Yihang She, Goutam Bhat, Martin Danelljan, Fisher Yu
IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2022
Digital Taxonomist: Identifying Plant Species in Community Scientists' Photographs
Riccardo de Lutio, Yihang She, Stefano D'Aronco, Stefania Russo, Philipp Brun, Jan Dirk Wegner, Konrad Schindler
ISPRS Journal of Photogrammetry and Remote Sensing, 2021

You can find the full list of my publications on Google Scholar.

Education

University of Cambridge
PhD in Computer Science
ETH Zurich
MSc in Geomatics · Recipient of the Excellence Scholarship
Nanjing University
BSc in Geographic Information Science