Sense Aeronautics’ AI-powered Automatic Target Recognition (ATR) solution automatically detects and classifies targets of interest in EO and IR video streams, delivering results that are faster, more cost-effective and more consistent than human operators. The system is highly scalable and can handle any number of video streams, making it ideal for drone fleet and swarm applications.
By integrating seamlessly into the mission workflow and accurately identifying relevant information, the system reduces storage and bandwidth requirements. Data concerning threats and events of interest can be output to external systems such as video management software, ground control stations and infrastructure monitoring systems.
The latest ATR solution enables the retrieval of telemetry data from drones enhancing AI performance and enabling the provision of geolocation data for detected targets, making it possible to display targets on a map.
ATR is available for deployment on the edge, as well as on local, public, and private Cloud-based environments.
Specifications:
| Input Sources |
USB and GbE cameras IP Cameras Video files (MPEG-4, H-264, etc) Video streams (RTP, RTSP, RTMP) Video platforms (Milicast, Dolby.io, DJI) |
| Output Formats | Dolby, RTMP, RTSP, WebRTC, HLS |
| Training Data | 150k images with over 2 million annotations |
| Latency |
<200ms for detection <1s for remote video availability on cloud deployment |
| Detection Probability | >90% |
| Detectable Target Size | <10% |
| Detection Range (Person) |
DJI Zenmuse H20 - 3000m DJI Zenmuse Z30 - 4500m Flir BlackFly (6mm lens) - 250m Mavic 3ET – Thermal Camera - 200m |
| Detection Range (Vehicle) |
DJI Zenmuse H20 - 5000m DJI Zenmuse Z30 - 8000m Flir BlackFly (6mm lens) - 400m Mavic 3ET – Thermal Camera - 300m |
| Deployment | Edge (onboarded), Local (GCS), private cloud or public cloud |
