Image courtesy Chess Dynamics
 
DEFT utilises a deep learning approach, developed by Chess brand Vision4ce, to create a comprehensive model of the tracked target, allowing the system to accurately locate dynamic targets and reliably re-acquire the target following periods of occlusion. DEFT provides improved tracking of difficult targets, such as dynamic objects that are changing appearance or rapidly accelerating, against foreground and background clutter, where traditional algorithms struggle.

 
The technology enhances Chess’s AI-driven target detection and tracking capability and integrates with Neural Network-based object detection and classification of targets including multi-rotor drones, vessels and land vehicles.
 
As the tracking progresses, the model is continuously fine-tuned to enhance its understanding of the target, resulting in precise long-term, robust tracking performance. Accurate classification allows the system to take advantage of automation technology for threat prioritisation and alerts, while also developing robust analytics data to enable more accurate decision making.
 
David Tuddenham, Chess Group Managing Director said: “The increased proliferation of stealthy drones and more flexibly deployed forces has posed an unprecedented threat to security and privacy. These agile and hard-to-detect devices capitalise on cluttered environments to evade traditional surveillance methods, highlighting the urgent need for innovative technologies to counteract them.
 
"DEFT has been developed in response to this growing issue. It exhibits remarkable resilience in the face of dynamic targets, effortlessly adapting to changes in appearance or rapid acceleration. Adopting Vision4ce’s highly efficient technology will allow us to address a critical challenge in modern surveillance.”