Currently, various satellites are used to monitor human effects on the environment. Satellite imaging is necessary for tracking the Earth’s characteristics, controlling assets, and analyzing impacts on the environment. Accurate and reliable Land Use Land Cover (LULC) details are necessary to evaluate ecological conditions and control resources from nature. However, it is hard to collect adequate time-series data with excellent spatial and temporal resolution using a single sensor. The lack of training information limits classification efficiency, particularly when dealing with data imbalance issues during spatial analysis. In order to tackle multiple complications in the existing technique, a new LULC classification mechanism is implemented with the deep learning technique. In the beginning phase, the required satellite images were collected from the standard database. Next, the LULC classification procedure is performed through newly designed Attention-based Squeeze-and-Excitation Networks (A-SENet). Finally, land use and land cover classification results are obtained from A-SENet. Later, multiple observations are performed in the suggested technique by considering various performance measures over the existing mechanism. In the accuracy validation, A-SENet accomplishes higher efficiency as 94.32% for dataset 1 and 93.7% for dataset 2 by considering batch size 64. Thus, the validation outcomes displayed that the suggested framework is more suitable to use with various applications.