Clouds can provide information about the weather. Efficient climate forecasting on precise rainfall prediction can assist in suitable agricultural planning. In order to forecast rainfall using cloud images, a novel method based on transfer learning techniques is presented in this study. Through identifying different kinds and states of clouds in the sky, the proposed work seeks to offer important insights on predicted patterns of precipitation. In this work, transfer learning techniques are applied to refine the model with three pretrained convolutional neural network models like VGG16, Inception-V3, and ResNet. The comparative analysis shows that transfer learning streamlines the model’s performance to a great extent. The results are compared, in terms of accuracy, precision, and recall to obtain the best among them.