A Learning on Intensity Prediction of Tropical Cyclone Infrared Images by Gabor Filter on Binary and Multi Class Approach

Tropical cyclone intensity classification is critical for disaster preparedness and resource allocation. Existing methods rely heavily on either manual analysis or computationally intensive deep learning models, which, despite their high accuracy, are often impractical for real-time scenarios. A significant gap exists in the literature where lightweight yet accurate models optimized for real-time applications are underexplored. This study addresses this gap by leveraging Gabor filter-based texture feature extraction combined with machine learning models, enabling precise cyclone intensity classification while balancing computational efficiency and prediction accuracy. By evaluating multiple classifiers, including Random Forest, SVM, and KNN, this study offers a comparative perspective to identify the most effective model for cyclone intensity classification in binary and multi-class setups.