Edge-cloud Collaborative Electric Power Inspection Image Target Detection Method

Abstract

Electric power companies use UAVs, cameras, robots and other equipment for electric power transmission line inspections and image collection. Further, through artificial intelligence image detection technology, the images are processed to identify the defects or hidden dangers. Image detection model services are generally deployed on cloud platform, which is the cloud, or edge computing devices, which is the edge. On the one hand, if all images are processed at the edge nearby, the calculation will take a long time due to the limitation of edge computing capability. On the other hand, if all images are uploaded to the cloud for process, it will still take a long time when the bandwidth is limited. Transferring too much data will lead to inefficiency, and it is unrealistic to upload all data to the cloud. Edge-cloud collaborative electric power inspection image target detection method splits and deploys the image detection model services to the cloud and the edge. They are all involved in calculation, which reduces the amount of data transmitted and improves efficiency.