Comparative Analysis of Pre-Trained CNN Architectures for Multi-Class Weather Classification: A Study on Deep Learning Techniques

In recent years, convolutional neural networks (CNNs) have proven to be highly effective in a wide range of computer vision tasks, including image classification. In this study, we compare the performance of several pretrained CNN models on the task of weather classification from single images. We evaluate the models on a dataset of weather images, consisting of five classes: cloudy, foggy, rainy, shine and sunrise; We consider several popular CNN models, including MobileNet V2, Inception V3, Densenets 121, 169 and 201 and finally the Xception model which were all trained on the ImageNet dataset.We conduct experiments to evaluate the performance of each model on the weather classification task.