Research on marine litter identification model based on convolutional neural network

The problem of marine litter is becoming more and more serious, which poses a great threat to the marine ecological environment. The research of marine litter identification and classification problem is imminent, and the identification and classification of images are especially critical. The advantage of convolutional neural network, one of the deep learning models, lies in its ability to automatically learn the features in the image, thus improving the accuracy of image classification. Therefore, the study of marine litter recognition model based on convolutional neural network is of some significance. Collecting and pre-processing multiple types of marine litter image data, extracting features in terms of colour, shape and texture, constructing a marine litter recognition model based on convolutional neural network, training and analysing the comparison results, effectively verifying the accuracy and precision of the model. The proposed model has high accuracy and stability in marine litter identification, which can provide effective technical support for marine environmental protection as well as a technical basis for the protection of marine ecosystems.