Resnet-50 Integrated with Attention Mechanism for Remote Sensing Classification

Abstract

Land use and land cover (LULC) classification is essential in many areas, including environmental monitoring and urban planning. There is an increasing need to improve classification accuracy. The existing approaches like VGG16, AlexNet, and ResNet50 have been widely used for this task. A novel strategy is put forth to tackle this problem that combines the ResNet50 architecture's strength with a channel attention mechanism (CAM). By carefully emphasizing the appropriate image elements, this novel technique seeks to increase accuracy while facilitating more accurate LULC categorization. By adding CAM to ResNet50, the model gets better at spotting complicated patterns and minute differences within land use land cover categories, leading to better classification outcomes. The EuroSAT dataset has gained extensive popularity in land use and land cover (LULC) classification endeavors, providing valuable support to researchers by facilitating the training and evaluation of machine-learning models for the precise categorization of satellite images. The outcome of this study indicates that the proposed methodology performs better than conventional methods, providing a promising route for develo** image classification techniques, particularly in the context of LULC classification, where accuracy is crucial for well informed decision-making and environmental management.