AI VisionCube D

The AI VisionCube D combines a wide-field short-focus visible camera with a 12mm telephoto module on a single 6 TOPS image-processing board, giving drone operators full situational awareness and high-zoom target detail in the same field of view. Built around a deeply-optimized YOLOv7 architecture, the AI VisionCube D detects up to 50 dynamic targets per frame and locks onto human and vehicle subjects at ranges beyond 1,200 meters — over 85% higher recognition accuracy than conventional single-camera AI modules.

Onboard intelligent crosshair adsorption pulls tracked targets to the center of the frame in 0.5 seconds, dramatically reducing the micro-control workload on the pilot. In high-speed motion scenarios — moving vehicles, evasive subjects, dynamic environments — the lock success rate reaches 92%. When targets are briefly occluded by terrain, foliage, or other obstacles, Kalman-based trajectory memory re-acquires them within three seconds and resumes stable tracking.

Engineered for plug-and-play integration, the AI VisionCube D supports BetaFlight and ArduPilot via the CRSF protocol, drops into 7-inch and larger industrial UAV airframes in under five minutes, and runs on a 9–16V input with full analog video output. End-to-end video latency stays under 60 milliseconds, enabling smooth real-time decision-making across long-range tracking missions. Total module weight comes in below 200 grams including all camera modules.

Six remote-control channels manage the full feature set: target lock, snap-tracking enable and disable, fine-adjustment crosshair movement, camera switching between short-focus and long-focus views, and self-stabilization mode toggling. Picture-in-picture mode displays the full-scene short-focus view in the main frame while overlaying 8× zoomed long-focus detail in the sub-view, letting pilots maintain situational awareness while tracking specific targets at maximum zoom.