Research on Rural Landscape Feature Recognition and Classification Based on Machine Learning

With the acceleration of urbanization, the protection and research of rural landscape is particularly important. Traditional rural landscape feature recognition methods rely on manual observation, which is inefficient and easily influenced by subjective factors. This study uses machine learning technology, especially the convolutional neural network (CNN) in deep learning, to improve the accuracy and efficiency of rural landscape feature recognition. By constructing a transfer learning model based on VGG16 and proposing a multi-scale feature fusion CNN (MSF-CNN) model, this study has achieved high-precision identification and classification of rural landscape features. A large number of rural landscape images were collected by combining aerial photography with ground photography, and the images were carefully marked and preprocessed. The performance of the original CNN model, the fine-tuned VGG16 model and MSF-CNN model in the task of rural landscape feature recognition was studied and compared. The results show that MSF-CNN model has achieved the best performance in accuracy, recall, precision and F1 score, which proves the effectiveness of multi-scale feature fusion strategy. In addition, the stability and generalization ability of the model are verified by K-fold cross-validation method. In this study, a new method of rural landscape feature recognition and classification is proposed based on machine learning technology, and its effectiveness is verified by experiments. This method not only improves the accuracy of identification, but also provides a scientific basis for the protection and development of rural landscape.