항 목 | 특 징 | 비 고 |
---|---|---|
GigE Vision Streaming | Video Streaming Input and Output over GbE | |
GPU | NVIDIA Pascal 256GPU (1.3 TFLOPS) | 1x 2.0 HDMI Type A |
CPU | 6-core ARM CPU | |
Memory | 8GB | |
Programming Language | Python | |
Libraries | Open CV, TensorFlow, TensorRT, CUDA | |
Ethernet | 5x Gigibit Ethernet ports (4x PoE 30W PSE IEEE802.3at) | |
USB | 4x USB 3.0 Type-A, 1x USB 2.0 Micro-B OTG | |
Serial Communications | 1x RS-485, 1x CAN 2.0b | |
Size | 226.0 x 122.0 x 59.9 mm | |
Temperature | Operating (0℃~70℃), Storage (-20℃~85℃) | 40℃ at 95% Humidity, Non-Condensing |
Power | In (54V/2.78A), Consumption (15W) | NVIDIA System on Module without PoE |
01Define classes of interest
02Automatic model-search
03Improve labeling
04Fine-tune performance
항 목 | 요 구 사 항 | 데이터 포맷 |
---|---|---|
perClass Mira GUI | MS Windows 7-10, Linux 64-bit | ENVI(BIL, BIP, BSQ layouts, uint8, uint16, float, double precision data) Matlab .mat files Tiff files, one for each spectral bands |
Optional NVIDIA GPU with CUDA9 and later or Open CL | ||
perClass Mira Runtime DLL | Multi-core CPU and GPU acceleration | |
MS Windows 64-bit (x86 and NVIDIA Jetson) |