Edge AI Suite
Advantech has integrated OpenVINO (Open Visual Inference & Neural Network Optimization) and CUDA (Compute Unified Device Architecture) toolkits into Edge AI Suite 2.0. Therein OpenVINO was released by Intel and is able to execute real-time inference for collected data on Intel CPU, GPU, FPGA or ASIC (IPU or VPU-Moviduis) acceleration chipsets to upgrade computing power. Its Model Optimizer module can translate other common AI frameworks, such as Caffe, Tensor flow, MXNet and ONNX, to a unified architecture for OpenVINO inference computing. Similarly, CUDA is adopted to realize AI inference demos on NVIDIA GTX/RTX graphic cards and Jetson (SOM) series. Regarding OS support, Edge AI Suite can run on Windows 10 or Ubuntu 18.04/20.04.
Quick Start on Intel / nVidia
This page demostrates how to launch an inference program, following the steps to choose one application you want to activate, video source and acceleration chipset you want to use and you wil see an AI inference running in another window.
It shows payload and temperature for each chipset in this page, and by clicking on “Run” button we can evaluate each chipset performance and FPS results.