Implementation Of Computer Vision Model In Edge AI Based By Using Intel Distribution OpenVINO Toolkit

In many technical fields of computer vision,target detection is a very basic task.Image segmentation,object tracking and key point detection usually depend on target detection.At present,a large number of edge AI devices have been equipped with cameras.In order to process the captured images,edge AI devices need to have the ability of target detection.At present,most of the mainstream high-performance target detection technologies use complex deep learning models such as convolutional neural network.Using edge devices to perform these tasks will face the problems of insufficient computing power,long time delay and high energy consumption.Therefore,in the computing scenario of edge AI devices,in order to execute complex target detection models on edge devices in real time,the complex models must be compressed and optimized.Based on this,this thesis designs a real-time target detection system suitable for edge AI equipment.The main work and contributions of this paper are as follows:1)this thesis provide a comprehensive overview of the challenges that exist in present computer vision applications,as well as Open VINO toolkit and Artificial Intelligence(AI)on Edge as potential solutions.2)By redesigning the methodology for deploying computer vision(CV)systems,the author provided an efficient way to construct and deploying edge applications.Research shows that this edge-node deployment technology will provide high performance processing capability for edge-node installations of actuators,sensors and iot devices.