Transfer Learning Approach - An Efficient Method to Predict Rainfall Based on Ground-Based Cloud Images

Clouds play a vital role in climate prediction. Rainfall prediction also majorly depends on the status and types of clouds present sky. Therefore, cloud identification is most exciting topic meteorology attracts researchers from other areas. This paper presents transfer learning technique to predict based ground-based Cloud images responsible for rains. It will estimated by identifying type taking as input. The dataset are divided into three categories(classes) labeled no-rain very low-rain, low medium-rain, medium high Rain associated Precipitation appropriate Rainfall. model be helpful farmers manage their Irrigation knowing before every irrigation cycle or can take decisions outdoor events prior knowledge Rain. trained classes firstly experimented with CNN. To improve performance, experiment carried out some best-pretrained models VGG16, Inception-V3, XCeption using and, results compared regular CNN model. outperformed get good accuracy too small presented best possible Google colab GPU setting makes task fast efficient time, performance achieved excellent fulfill real-time requirements.