Identifying weather conditions is a prevalent issue across all industries. Autonomous vehicles must accurately assess meteorological conditions to adapt their driving behavior. Contemporary farming relies on analyzing present weather patterns. A potential approach could involve implementing a system that detects weather conditions from images. Since no specialized sensors are required, the entire setup should be quite affordable. Transfer learning (TL) enables the development of picture categorization methods with a limited database. This work introduces 3 weather detection systems. The designs are built using InceptionV3, MobileNetV2, and ResNet50 designs. Their effectiveness is evaluated and explained. © 2024 Elsevier B.V., All rights reserved.