Underwater Image Processing and Target Detection Algorithm Based on Deep Learning

Abstract

AbstractWith the growth of the earth’s population, people’s demand for space and resources continues to grow, making marine environmental research and exploration a new field that human development needs. The composition of the marine environment is unique to humans, and the application of intelligent underwater vehicle equipment with the ability to autonomously detect and recognize the environment can expand the scope of human research in the ocean. This article focuses on the research of underwater image processing and target detection algorithms based onDL, and understands the relevant theories of underwater image processing and target detection based on the literature, and then the underwater image processing and target detection algorithms based on DL In the design, in order to improve the accuracy of target detection, the DL model is optimized, and the optimized model is tested. The test result shows that if the Atrous convolution operation is added, the calculation amount of the network model will increase, resulting in detection.The speed drops to 50FPS. But after adding Atrous processing, the detection accuracy will be improved accordingly.KeywordsDeep learningUnderwater imagesImage processingTarget detection