At the heart of the DS071 lies the RV1126B AI vision processor, a quad-core ARM Cortex-A53 chip running at 1.6GHz. Its built-in NPU delivers up to 3TOPS of computing power, enabling real-time target tracking at an impressive 120Hz frame rate. This high refresh rate ensures that fast-moving objects remain locked with ≤1 pixel tracking accuracy. In addition, the board supports multiple tracking algorithms, including centroid tracking, correlation tracking, multi-target tracking, and intelligent tracking, adapting to different target types and scenarios. For detection, users can choose from AI recognition, centroid search, template matching, and moving target detection. Consequently, the DS071 excels in complex environments where both stationary and moving objects must be tracked simultaneously. Beyond tracking, the board offers flexible video I/O: two MIPI CSI inputs, one USB 3.0 input, CVBS analog output, and a 100M Ethernet port for networked streaming. Video compression is handled efficiently with H.264 and H.265 encoding, and streams can be output via RTSP, UDP, or RTMP. Moreover, the DS071 provides a rich OSD overlay capable of displaying time, date, target boxes, crosshairs, scale, compass, and even geolocation data like latitude and longitude. This overlay is invaluable for UAV target tracking missions where operators need comprehensive situational awareness. Power management is another strong suit. The board consumes less than 4W at a nominal 12V DC, making it ideal for battery-powered platforms. Weighing only 20.1 grams (bare board) and measuring just 43mm × 38mm, the DS071 can be seamlessly integrated into space-constrained gimbals and small UAVs. Furthermore, the optional development board version provides an Ubuntu environment with SSH access, SDK resources, and API documentation, giving developers full control over custom algorithms and application logic. All these features make the DS071 a truly versatile AI video tracking board for edge-based vision applications.

The DS071 is purpose-built for UAV target tracking and other mission-critical embedded vision systems. On unmanned aerial vehicles, the board’s ability to automatically detect, lock, and track ground or aerial targets in real time significantly enhances autonomous operation capabilities. For instance, it can track moving vehicles for traffic monitoring, follow boats for maritime surveillance, or identify and track intruder drones for counter-UAS applications. In gimbal payloads, the DS071 adds intelligence to EO/IR cameras by enabling automatic target recognition and continuous tracking, freeing operators from manual control and reducing cognitive workload. Beyond airborne platforms, this compact AI video tracking board excels in ground-based embedded vision systems. It can be deployed in fixed or mobile surveillance stations for perimeter monitoring, automatically scanning for targets and triggering alerts when anomalies are detected. Its support for visible-light and infrared dual-channel input makes it suitable for all-weather security. Moreover, the development board version opens new possibilities for research institutions and AI startups: with Ubuntu, SSH, and rich SDK documentation, engineers can prototype custom neural network models and test novel tracking algorithms directly on the hardware. This flexibility accelerates the development of specialized embedded vision systems for industrial inspection, autonomous robotics, and smart city applications. In summary, the DS071 bridges the gap between advanced AI tracking algorithms and real-world deployment, delivering a reliable and cost-effective solution for a wide spectrum of vision-based automation tasks.