Cyclones are enormous storms that bring heavy rain and strong winds. The INSAT 3D satellite accurately tracks cyclones and their progression. The objective of our work is to measure the severity of the cyclone using the generated sequence of photographs. Estimating cyclone strength is essential for timely alerts, risk analysis, emergency management, and understanding cyclone behaviour. It helps with planning, resource allocation and minimising the effects on people’s lives and infrastructure. Current cyclone intensity detection systems lack complexity and dynamics, leading to inaccurate predictions and ineffective reaction measures. Improved strategies for detecting cyclone intensity can improve catastrophe preparedness and response in cyclone-prone locations. This work is centered on the study of cyclone intensity prediction using INSAT-3D satellite images. INSAT stands for “Indian National Satellite System”. Utilizing historical data and atmospheric features captured by INSAT-3D, we employ deep learning models to capture complex temporal correlations. These architectures excel in anticipating cyclone strength, facilitating timely alerts and effective emergency management. The comparative analysis demonstrates that our proposed Convolutional Neural Network model performed better over existing methods. By embracing cyclone dynamics, we provide precise intensity assessments, enhancing catastrophe preparedness and response.