Transfer Learning-Based Sustainable Climate Recognition System

A climate recognition system (CRS) based on transfer learning (TL) technique has been proposed for multi-classification of collected climatic images based on 4 different climatic classes. MobileNetv2 model along with a simple CNN classifier has been fusion up to develop an efficient and sustainable CRS for recognition and classification of various climatic conditions in different terrains. Rainy, sunny, foggy, and windy climatic classes have been considered for multi-classification which resulted in the best recognition accuracy of 98.75
^%
in the case of the sunny climatic class. The proposed work has also discussed the methodological milestones adopted for the conduction of the experiment, with prominent challenges encountered. Additionally, the proposed technology contributes to the provision of sustainable support for education and quality of life. The observation and findings of the proposed work have proved the efficiency, sustainability, and efficacy of the TL approach for developing CRS.