Shaogang Gong is a Professor of Visual Computation at Queen Mary University of London and the Head of the Computer Vision Group at QMUL. He is a Fellow of IEE (now IET), a Fellow of the British Computer Society, and a Member of the UK Computing Research Committee. His research is in computer vision, machine learning and video analysis, with applications in visual surveillance and video forensic analysis. He has developed mathematical models, algorithms, patents, and commercial software for video behaviour recognition and multi-object multi-camera tracking, in particular person re-identification.
He is interested in variational graph models including Bayesian graphs, dynamic Bayesian networks, hidden Markov models, and probabilistic latent semantic analysis. His recent work includes domain transfer learning, unsupervised and semi-supervised deep learning, imbalanced data deep learning, reinforcement learning, zero-shot learning, and human-in-the-loop active learning.
The amount of video data of public urban space is growing exponentially from 24/7 operating urban camera infrastructures, online social media sources, self-driving cars, and smart city intelligent transportation systems, e.g. 1.4 trillion hours CCTV video in 2017, growing to 3.3 trillion hours by 2020. The scale and diversity of such video data make it very difficult to filter and extract useful critical information in a timely manner. There is a fundamental challenge to develop data analysis tools for large scale video semantic search by exploring the huge quantity of video data using deep learning, critical for smart city on public security, safety and intelligent transport.
To that end, robust and scalable algorithms are required for human recognition, of both individuals and populations, and searching individual people and/or categories of people and objects (semantic search) in large scale video data from diverse multiple sources city wide. This Turing research project aims to develop both scalable algorithms and software for semi-supervised and unsupervised deep learning for domain transfer object search, attribute-based latent semantic embedding space inference, and deep learning knowledge distillation.