Jinwei Gu
Papers
4
Total Citations
261
H-Index
4
About
Jinwei Gu is a leading researcher in computer vision and computational imaging, best known for his pioneering work in 3D scene understanding and image restoration. His most impactful contribution is PlaneRCNN, a deep neural architecture that detects and reconstructs piecewise planar surfaces from a single RGB image. By adapting Mask R-CNN to jointly predict plane parameters and segmentation masks, Gu’s method enables robust 3D plane detection and reconstruction, earning over 240 citations and setting a new standard for single-image geometry recovery. Beyond 3D vision, Gu has advanced computational photography through his work on pinhole photography restoration, applying modern machine learning to low-light denoising, HDR imaging, and demosaicing. He has also contributed to depth completion challenges, combining RGB images with sparse Time-of-Flight measurements for robotics and autonomous systems. Gu’s research bridges the gap between traditional optics and deep learning, producing practical solutions for consumer and mobile photography. His work on PlaneRCNN remains a cornerstone in the field, demonstrating how neural networks can extract rich 3D structure from minimal input. With a strong record of high-impact publications and active participation in benchmark challenges, Gu continues to shape the future of intelligent imaging and scene understanding.
Research Focus
Key Achievements
Top Papers
- 1PlaneRCNN: 3D Plane Detection and Reconstruction From a Single Image240 citations · 2019
- 2
- 3MIPI 2023 Challenge on RGB+ToF Depth Completion: Methods and Results6 citations · 2023
- 4PlaneRCNN: 3D Plane Detection and Reconstruction from a Single Image6 citations · 2018