Reinforcement & Deep Machine Learning Case Studies
Tracking a Watch List
Police forces worldwide often need to track a watch list of potential criminals or terrorists. It is not effective for manual surveillance given the cost and number of trackers required for every target. Facial recognition doesn’t work in real world challenging conditions. VSL has developed a system which integrates RE-ID with facial recognition, so that a person can be found in video footage even if there is no clear facial image or the lighting is poor or the person is partially occluded. The VSL system is scalable to large population of millions of people. Contact us to discuss the case studies where Police forces globally have deployed our systems.
Responding to Trigger Events
Cities around the world are transforming to smart cities where they exploit the benefit of video to improve public safety and transport. Typically, CCTV video footage is sent to a control room where there are watchers trying to respond to a trigger event. However, often the event is missed. VSL has AI algorithms with behaviour recognition capability to be able to identify predefined trigger events in real time. The system will provide an alarm for a predefined trigger event and then track it in real time to enable the controller to provide real time situation awareness to the support team on the ground.
In public safety, the trigger event can be defined as criminal behaviour, social disturbance or a possible terrorist event. In public transport, the triggers are mostly defined to enable controllers to maintain transport efficiency but can also cater for disruptive events like possible suicide on the track. Contact us to discuss the case studies where our solutions have been deployed in Smart Cities.
Crowd Control and Social Distance
Crowd analysis shows what is happening to an entire walking population, in real time in a variety of environments. This enables infrastructure operators like rail, airports, shopping centres and sports stadiums to take positive actions to increase efficiency but maintain public safety.
The COVID-19 Pandemic has enormous health and economic impact and one mitigation tool to contain the virus is work restrictions social behaviour guidelines. As the spread and impact to front line health care improves ease of restrictions will occur but will require further tools to mitigate an increase of viral spread. One of these tools will be the ability to be able to detect large crowds and classify their behaviour to help mitigate them forming in the future.
VSL deploys its expertise in Visual Analytics and Deep Learning Networks to apply Re-Identification of people (Re-ID) to Crowd Analysis. From a Video it can be determined if a Crowd forms, the density within the Crowd (Social distancing) and the movement This can then give an event trigger to the authorities on the location and crowd characteristics so action can take place to help mitigate future events. It is also noted that Re-ID only determines which people are different within a viewed crowd with no central data base with personal identification taking place (privacy is maintained).
Please feel free to contact us to discuss any of our case studies where we have helped to enhance surveillance solutions by deploying our AI solutions for Crowd Management.