Stephen Lin
Also known as Steve Lin
Stephen Lin is a computer vision researcher at Microsoft Research Asia whose work spans computational imaging, visual recognition, and generative vision. He holds a B.S.E. in electrical engineering from Princeton and a Ph.D. in computer science and engineering from Michigan, and he joined Microsoft Research Asia in June 2000. His papers include work on image reflection separation and depth from defocus, Deformable ConvNets V2, and Swin Transformer, which won ICCV's 2021 Best Paper Award, the Marr Prize. In 2025, he coauthored VASA-3D, an audio-driven 3D head-avatar generator. As of October 2026, he is a Senior Principal Research Manager in Microsoft Research Asia's Visual Computing group.
- Now at
- Microsoft Research
- Works
- 185
- Citations
- 56,392
- h-index
- 64
Publication numbers from OpenAlex, as of 15 October 2026.
Timeline
-
2000
Joining Microsoft Research Asia
His degrees are a B.S.E. in electrical engineering from Princeton and a Ph.D. in computer science and engineering from Michigan. He joins Microsoft Research Asia in June 2000.
-
December 2001
Separating image reflections
He and Heung-Yeung Shum publish their CVPR paper on separating diffuse and specular reflection in color images in December 2001.
-
December 2010
Depth from defocus
With Changyin Zhou and Shree K. Nayar, he publishes 'Coded Aperture Pairs for Depth from Defocus and Defocus Deblurring' in the International Journal of Computer Vision in December 2010.
-
2019
Deformable ConvNets V2
He coauthors Deformable ConvNets V2 at CVPR in 2019. Its revised form focuses features on relevant image regions and improves results on COCO for object detection and instance segmentation.
-
2021
Swin Transformer
He coauthors Swin Transformer in 2021, a hierarchical vision Transformer that limits self-attention to local windows and allows connections across them.
-
2021
The ICCV Marr Prize
Swin Transformer wins ICCV's 2021 Best Paper Award, the Marr Prize.
-
October 2025
Audio-driven 3D avatars
In October 2025, he coauthors VASA-3D, an audio-driven generator that builds a 3D head avatar from one portrait image.
-
As of October 2026
Senior Principal Research Manager
As of October 2026, he is a Senior Principal Research Manager at Microsoft Research Asia and a member of its Visual Computing group, where he studies computer vision.
Sources and credits
Photo: We have no published photo of Stephen Lin yet. Send us a free-licensed photo.