Intel OpenVINO Toolkit for Computer Vision: Object Detection and Semantic Segmentation

Abstract

The paper provides an overview of the neural networks implementation current state, their methods of execution, and the Intel® OpenVINO ™ Toolkit for executing neural networks on various hardware platforms from Intel. This work describes the selection of computer vision neural networks and datasets for object detection and semantic segmentation of images for subsequent testing. It gives the description of the experiment on various hardware platforms. Moreover, it provides an analysis of the performance and cost of running selected neural networks using the OpenVINO ™ Toolkit in normal mode, as well as using plug-ins for multiple devices and heterogeneous plug-ins on multiple connected devices.