Register for free to
view the full innovation profile.
Platform-agnostic backend software that uses vision-based AI and sensor fusion to provide up to 360° of enhanced situational intelligence on moving camera platforms.
About
FortifAI is a platform-agnostic software application that uses vision-based AI and sensor fusion to provide 360° of enhanced situational intelligence on the edge without the internet, even on fast-moving platforms. When small, barely discernible to the human eye, objects are detected, FortifAI provides precision camera platform control to allow more granular target object analysis, such as the colour, behaviour, and visible written marking. Notable projects: FortifAI was first trialled in 2021 on board a Royal Navy unmanned submarine as part of a DASA contract. FortifAI downloaded the last known positions of "dark vessels" from an AI satellite data provider, visually classified other vessels via the periscope and used real-time AIS reports to confirm the ships were transmitting the correct class. We are the Detection Lead in a £12M consortium with other companies such as BT to build the "world's longest drone flight corridor". As part of this two-year Innovate R&D project, we must prove reliable 4KM visual detection of 4mx2m aircraft in real-time using non-zooming cameras. We have been awarded a 6-month DASA contract to further develop our airborne 360° Detect and Avoid with cooperative ground-based camera tracking. In November, we will demonstrate airborne drone detection and tracking on a fixed-wing drone as part of a Honeywell Aerospace paid pilot.
Key Benefits
Easy integration and no multi-AI supplier dependency With a single software integration, our customers empower their cameras with multiple capabilities, typically involving various AI providers. All capabilities can be used simultaneously and include the following: multi-domanial object detection and tracking, ANPR, colour recognition, OCR, distance estimation, sensor fusion, behaviour analytics, ArUco marker tracking, and face redaction. Platform Agnostic ForitfAI can be run on Windows or Linux on edge, on-premises or in the cloud. Using NVIDIA Jetsons, we have installed it on UAVs, UGVs, UUVs and within long-range EO/IR cameras. Precision camera platform control Although the importance of camera control can often be underestimated, entirely automating all camera movement significantly increases the speed that objects of interest can be found and kept within the camera view. Our software includes automated camera tests to characterise the mechanical and video response of each system it controls to ensure optimal performance. We are working on extending the platform control to drones too, where we control the drone for specific tasks such as take-off and landing on a moving vehicle/vessel. Per person risk index FortifAI allocates a risk index to each person that is tracked. This can trigger alerts based on behaviour and prioritise which person is perceived as a higher threat. 360° Detection and Tracking Via Innovate UK and DASA funding, we are building very long-range object detection that can be used to detect and track air traffic from the ground and from the air. On the ground-based system, our goal is to detect 4m x 2m objects with a fixed view lens out to a maximum of 4KM with simultaneous processing of 8 video streams. The airborne version includes image stabilisation for more robust tracking and uses a single model for both detection and tracking to maximise efficiency when attempting to track objects as small as 8 pixels. Detection does not rely on a Moving Target Indicator, meaning stationary threats that would have been missed by more traditional video tracking systems.
Applications
Uses cases include: -Maritime vessel detection, tracking and visual confirmation of AIS-reported vessel type -Detect and avoid enabling drones to fly BVLOS safely without collision -Ground-based or airborne C-UAS to reduce false alarms from external sensors such as Radar or RF and reduce the operator's cognitive load with automated drone search -Counter-terror and suicide prevention using per-person or vehicle behavioural analytics -Extended surveillance (i.e. Special Forces) by logging person and vehicle activities.