Why Vision Semantics
Artificial Intelligence deployed in computer vision analytics is coming of age as the hardware is catching up with the algorithms and fast processing demands of the software. Vision Semantics is the driving force in this market with a 12-18-month sustainable technology lead in Re-ID over the nearest competitor. The technology is backed by 6 international patent families. This is combined with 10 years of commercial work with international Government and law enforcement agencies to ensure the technology is accurate, fast and scalable for real world deployments. Vision Semantics was created by Professor Sean Gong the world authority on unsupervised Re-ID and human supported machine learning, who leads the academic field in Person Re-ID. With a decade of operational commercial deployments within Vision Semantics to refine the state-of-the-art AI and deep learning, no other company is close to emulating this effort
World Class Research Driven Approach
The Queen Mary Computer Vision Group has been conducting world-leading research in computer vision and machine learning for over 20 years, and is internationally renowned for its work on video behaviour and action recognition, person re-identification, multi-camera tracking, and face analysis. The group was founded and is led by Professor Sean Gong, co-founder of Vision Semantics.
The core expertise of the group includes deep learning for generic visual computation, statistical modelling for pattern recognition, algebraic methods for visual information processing, biologically inspired mathematical modelling, zero-shot learning for scalable visual recognition, unsupervised learning for data structure discovery, human-in-the-loop modelling and reinforcement learning for adaptive model optimisation.
The main areas of application include but not limited to person re-identification, face recognition, action and behaviour analysis, clothing analysis, sketch recognition and synthesis, vehicle analysis, identity recognition, video summarisation, smart camera network, target detection and tracking, privacy, multi-view scene reconstruction, eye-movement, and 3D display.
Vision Semantics takes this research and commercializes it so that is robust, scalable and can be implemented by partners in real world situations.
Vision Semantics' Technology Approach is Unique
Vision Semantics began a series of algorithm developments for Re-ID based on supervised, semi-supervised unsupervised, and reinforcement learning using a diverse ethnographic set of benchmark data taken from 12 global cities (21 locations). Vision Semantics have developed a unique patent protected Deep Learning and Reinforcement Active Learning algorithms for optimising domain transfer zero shot learning in Re-ID. This combined with a ‘Human-in-the-Loop’ Search & Learn results in a continuous improving system without the need for large scale labelled training data sets at all target domains. We call it Dynamic Search & Learn – the more you use it the better it gets.
Vision Semantics’ approach has led to a robust Re-ID implementation which is scalable, quick to deploy and it does not need vast sets of training data at every user application target domain. Its ability to dynamically search and learn sets it apart from any other approach