Design of Image Recognition Monitoring System of Hydrological Monitoring Station Based on Edge Intelligence

Abstract

In the safety monitoring of hydrological monitoring stations, the target detection algorithm can be used to avoid unnecessary trouble caused by artificial supervision through intelligent monitoring. However, the accuracy of target recognition and time - delay is always a contradiction in security monitoring. SSD target detection algorithm is the latest target recognition algorithm after Faster RCNN and YOLOv1 algorithm, combining the advantages of both. The algorithm is faster than the fast RCNN algorithm and has higher accuracy than the YOLOv1 algorithm. In this paper, a PSO intelligent algorithm based on hyperlight fast generic face detection and 1mb network convergence is proposed, called PSO-1MB. The device is deployed in or near the hydrologic monitoring station on the edge node server for calculation and processing. In this paper, the test whether the staff wear a hard hat as an example, using the Pytorch environmental framework, experiment simulation. Experimental results show that this model and algorithm can detect helmet features more accurately and quickly, and can better meet the engineering requirements.