Analytics to Observe the Customer Behaviour

Abstract

Through the application of transfer learning techniques and INSAT 3D satellite imagery, this research endeavors to create a system capable of discerning the intensity of cyclones. As one of the most destructive natural phenomena, cyclones inflict extensive property and human casualty losses. Effective disaster management and mitigation endeavors rely heavily on the precise identification and timely evaluation of cyclone intensity. INSAT 3D, a geostationary weather satellite, produces images of the earth's surface in high definition. The photographs adeptly portray the atmospheric conditions that are favorable for the detection of cyclones. The presence of clouds, noise, and various atmospheric disturbances poses a significant obstacle in accurately differentiating cyclones from these images. We present an innovative transfer learning approach for cyclone intensity detection in this research article. Transfer learning is an approach to machine learning that enables the adaptation of models that have been previously trained to solve problems in different domains. To derive pertinent features from the 3D satellite images of INSAT, we utilize convolutional neural networks (CNNs) that have been trained since their inception on extensive datasets. The classifier employs these characteristics in order to discern and classify the cyclones in accordance with their intensity.