Design of an Underwater Fish Recognition System Based on Intelligent Edge Computing

With the increasing attention paid to aquaculture in recent years, fish detection and identification play a key role in environmental monitoring of aquaculture and fishery development. Recently, underwater target recognition technology has become an important part of exploring aquaculture, and accurate identification of fish in turbid waters of aquaculture is the guarantee of fishermen’s economic benefits. With the continuous pursuit of light weight, low cost and high performance, it is of great significance to combine edge computing platform with underwater image processing to design a set of underwater image processing system with high performance and low cost based on edge computing platform for real-time detection of underwater multi-targets. In this study, yolov5 target detection algorithm is adopted, and a large number of data sets in different environments are collected through the network. According to the training requirements, the cloud GPU is rented to complete the data set training and testing on the pc side, and according to the system requirements, Design the scheme of edge computing platform. Based on intelligent edge, this system aims to realize underwater fish identification design. Raspberry Pie 4B is selected as the edge computing carrier and runs on Linux system. In view of Windows10 IoT With the release of, users can run Windows operating system on Raspberry Pie and use Win32 Disk Imager. Burn the raspberry pie system into SD card, and connect the display screen and keyboard to open the raspberry pie. Because of this article’s The design is real-time display, so it is necessary to install an external camera and an opencv calling camera. Complete real-time display. Combined with underwater multi-target detection algorithm and edge computing platform, an edge computing-based algorithm is constructed. The underwater image processing system requires that the recognition rate of the selected fish should reach 50The system performance is analyzed to verify that it meets the practical application requirements.