512 core Volta @1377Mhz + 64 Tensor Cores, computing performance 38TOPS

32G LPDDR4, 32G onboard emmc, expandable NVME SSD hard drive

1000BASE-T Ethernet

USB3.0 interface

HDMI interface

Wide voltage input: 12-19V

Machine power: 35W

Working temperature: -25℃-65℃

IP64 protection class

CUDA10.2 + CUDNN8.0 acceleration library

TensorRT 7.2 acceleration engine, supporting 50FPS YOLO 602 16-bit floating point inference capability

Ubuntu18.04, GCC8.0 operating environment.

Application Scenario:

Real-time detection and tracking of industrial open-air visual targets, real-time detection and segmentation of medical visual targets, real-time target detection of forest fire prevention, real-time target recognition and early warning of geological disaster targets, and real-time detection and early warning of environmental disaster targets.