Marine Target Detection Method based on Embedded Vision

We propose to study the key technologies of marine ecological environment information acquisition and marine information visualization based on edge computing technology to improve the utilization of marine information resources. We apply edge computing technology to marine target recognition to solve the network bandwidth limitations it faces. Combining satellite message communication and embedded intelligent detection system to study marine information visualization model technology, and improve the monitoring ability of marine environment, marine resources, and marine specific targets. Combining with the deep learning framework, research on the method of sea and sea target recognition based on the embedded visual image data acquisition. Sea waves and rough sea surfaces are used as marine noise image samples, and objects such as ships, islands and reefs, and marine life are used as marine target image samples. In the work, a lightweight target detection framework network model is selected to train the samples to generate a marine target recognition network, which is deployed to intelligent terminal equipment to realize marine target detection and recognition.