Research on Water Quality Image Classification Method based on Convolutional Neural Network

Green water is the golden silver mountain, and water is the source of human life. The assessment of water quality in rivers and lakes, as well as the prevention of water pollution, is a continuous and arduous task. In recent years, Artificial Intelligence (AI) has experienced explosive development and has been widely utilized in various fields. Deep learning has achieved significant results in computer vision image classification. Therefore, this paper proposes a method for classifying water quality images based on a convolutional neural network (CNN) algorithm as the basis for improvement. The water quality dataset was collected with 5203 water quality images, and the training and validation sets were divided according to a ratio of 9:1, with 4685 images in the training set and 518 images in the validation set. After continuous improvement of the algorithm model and extensive model training, an average classification accuracy of 98.6% was achieved, demonstrating improved classification accuracy and effectiveness.