Vision4ce, a provider of rugged image and video processing solutions for unmanned systems, has developed true AI classification capabilities for its tracking products. The deep neural network-based classification application runs on Vision4ce’s embedded systems powered by NVIDIA’s Tegra architecture, and is ideal for any unmanned systems that require real-time video and image processing, tracking and classification detection technology.
Vision4ce’s deep neural network has been trained to localise and classify multiple objects on land, at sea and in the air, making it ideal for UAVs (unmanned aerial vehicles), UGVs (unmanned ground vehicles) and USVs (unmanned surface vehicles).
The system features the following capabilities:
- Low Latency
- Dual Processing Channels
- Object Classification
- Image Enhancement
- Object Tracking
- Sensor Fusion
- Video Compression and Recording – H.264, M-JPEG
- Electronic Image Stabilisation
- Video Streaming – RTSP / MPEG-TS
- Servo Platform Control
- Camera Control
- Panoramic Image Formation
With wide experience in real time video and image processing, Vision4ce can develop custom solutions for deployment on a wide range of unmanned systems. Their portfolio includes a line of GPU-based embedded platforms which are well-suited to AI edge computing.
The three videos below demonstrate Vision4ce’s deep neural network classification capabilities for air, land and sea targets: