Zexiang Xu

Papers

Total Citations

H-Index

About

Zexiang Xu is a leading researcher in computer vision and graphics, specializing in photorealistic scene reconstruction, inverse rendering, and material appearance modeling. His most impactful contribution is the **OpenRooms framework**, a groundbreaking approach for generating large-scale, photorealistic indoor scene datasets with accurate ground truth for geometry, materials, lighting, and semantics. This work, published in 2021 and already garnering 66 citations, democratizes dataset creation by transforming raw 3D scans into richly annotated, realistic environments—enabling advances in scene understanding, relighting, and neural rendering. Xu’s research directly addresses a critical bottleneck in the field: the scarcity of high-quality, labeled data for training deep learning models. By providing an end-to-end pipeline that is both accessible and scalable, his work has become a foundational resource for researchers working on inverse graphics, intrinsic image decomposition, and augmented reality. Xu’s contributions bridge the gap between synthetic data and real-world complexity, empowering the community to build more robust and generalizable vision systems. His OpenRooms framework stands as a testament to his vision of open, reproducible research that accelerates progress across computer vision and graphics.