Tao Xiang
Also known as T. Xiang, Tao A Xiang
Tao Xiang is a computer-vision researcher whose work links video analysis, person re-identification and sketch recognition. He served as principal investigator for an EPSRC project on abnormal public-space video, coauthored Sketch-a-Net and OSNet, and contributed to FashionViL and CLIP-based sketch retrieval. In 2024, he coauthored a CLIP-based method for sketch classification across unknown categories and abstraction levels. As of October 2026, he is a Distinguished Professor at the University of Surrey and Chief Scientist at Samsung AI Research Centre, Cambridge.
- Now at
- University of Surrey and Samsung
- Works
- 383
- Citations
- 39,594
- h-index
- 94
Publication numbers from OpenAlex, as of 22 October 2026.
Timeline
-
August 2009–October 2012
Abnormal video analytics
From August 2009 to October 2012, he serves as principal investigator for EPSRC project EP/G063974/1, which develops video analytics to detect abnormal behaviour in public spaces.
-
2015
A BMVC Best Science Paper
In 2015, he coauthors Sketch-a-Net that Beats Humans with Qian Yu, Yongxin Yang, Yi-Zhe Song and Timothy Hospedales; BMVC gives the paper its Best Science Paper award.
-
2019
OSNet for person re-identification
At ICCV in 2019, he coauthors OSNet, a lightweight person re-identification network that combines features from multiple visual scales. It uses 2.2 million parameters and surpasses human operators in re-identification accuracy.
-
2022
FashionViL for fashion data
In 2022, he coauthors FashionViL, a fashion-focused vision-and-language model built with two pre-training tasks tailored to fashion data.
-
2023
CLIP for zero-shot sketch retrieval
In 2023, he coauthors CLIP for All Things, which uses sketch-specific prompts for zero-shot sketch-based image retrieval at category and fine-grained levels.
-
2024
CLIP for human-drawn sketches
In 2024, he coauthors Do Generalised Classifiers really work on Human Drawn Sketches?, a method that adapts CLIP for classification across unknown categories and levels of sketch abstraction.
-
As of October 2026
Surrey and Samsung AI roles
As of October 2026, he is a Distinguished Professor at the University of Surrey and Chief Scientist at Samsung AI Research Centre, Cambridge.
Career
- Since 2020SamsungChief Scientist
- Since 2019University of SurreyDistinguished Professor
- 2021–2025Meta
- 2020–2023IFlyTek
- 2014–2016University of Electronic Science and Technology of China
- 2003–2016University of London
- 2003–2019Queen Mary University of London
Sources and credits
Photo: We have no published photo of Tao Xiang yet. Send us a free-licensed photo.
- Multi-Object Video Behaviour Modelling for Abnormality Detection and Differentiation
- BMVC 2015 awards
- Omni-Scale Feature Learning for Person Re-Identification
- Surrey AI research achieves world-leading technology for visual recognition of people
- Prof Tao Xiang | University of Surrey
- Do Generalised Classifiers really work on Human Drawn Sketches?
- Artificial Intelligence MSc | University of Surrey




