Our Story

Why we're here

We believe in augmenting human intelligence, not replacing it. We advocate AI as a tool to improve human productivity, not as a `lights-out’ black-box to replace human judgement. In 2005, Professor Sean Gong anticipated the growth of video data and initiated a multi-year research programme to apply machine learning to transform the capability of video analysis in complex urban environments. The goal was to take world leading research and create products that were scalable and reliable for the demands of the commercial world. That’s why we founded Vision Semantics in 2007.

Here is what we do

Imagine needing to find a missing child or a confused elderly person in a city like London. Where are they?  You are going to rely on the emergency services to locate them. Now public safety agencies have a responsibility to protect citizens and they have to find your loved one quickly so that harm cannot befall them.  But they also have to locate and track other people who have committed a crime and try to evade capture. Such agencies need to be ever vigilant and have the best tools available for timely action.

Video has provided a revolution in situational awareness with the ability to assess scenes remotely.  Arguably, it has been too successful.  We all carry a video camera in our phone, and video surveillance cameras are deployed in the hundreds of millions, with that number growing every year.  Video is ubiquitous. There is more video being created than can be possibly observed by human beings. Just 1-hour surveillance video from a city of 1+ million cameras (such as London) generates more data than the entire BBC TV archives since 1922.  But how can you find and locate a single person of interest and keep track of them, across busy urban areas from multiple dispersed cameras?  The task is immense and till no unmanageable.

Facial recognition is great in a controlled environment

How about tracking their face with facial recognition software?  Well, that doesn’t work on standard video footage where the size of a single face is usually less than 20x20 pixels. Facial recognition works on 300x300 worst case 60x60 pixels, full frontal, in good light, no motion blur, no sunglasses, no hats nor hoodies.   So facial recognition achieves less than 10% detection rates on normal video footage. This performance gets worse when the number of people searched gets larger.

In a vibrant city you need Person Re-Identification

So, you have to identify the whole body as it moves from camera to camera.  That is called person Re-Identification or person Re-ID. Vision Semantics has pioneered Person Re-ID over the last 10 years, with commercial deployments around the globe and using artificial intelligence to rapidly find people in time, with accuracy over large-scale crowded environments that public safety agencies operate in. Vision Semantics uniquely pioneered Re-ID with human-in-the-loop search & learn AI algorithms, bridging transfer learning, reinforcement learning, semi-supervised and unsupervised deep learning to enable rapid and scalable Re-ID in unknown target domains from an open-set world.

VSL Inside is the AlphaGo of Person Re-ID

The public demand their loved ones are found in time and that criminal behaviour is detected and enforced. You would want the best resources to find your loved one and quickly react and remove any threat to public safety. Public Safety by law has to find people in time and enforce the law.  AI powered Person Re-ID will be at the heart of the next generation video systems in smart cities.

Time matters. VSL inside AI that finds people in time.   
We built our company in the AI triangle between London, Cambridge, and Oxford.

We’re world class engineers driven from leading academic research, funnelled into an exciting start-up company. At our headquarters in London we’ve assembled a team that combines practical expertise in deep machine learning, algorithm development, visual data processing, rapid data processing, and commercial software development. Whatever their role, each of theteam combines
an uncompromising engineering mindset with an unwavering focus on executing and delivering the
best solution to service our partners.