Development of a UAV-based crop disease detection system using deep learning algorithms to enhance precision agriculture

The shift toward precision agriculture has made it clear that there is a need for smart and automated systems that can detect crop diseases promptly and with high precision. A research project is being discussed here, which presents a comprehensive drone-based solution for detecting cotton leaf disease, combining the latest advances in deep learning with the gradual training of the machine for rapid decision-making in agriculture. In the beginning, the system implements adaptive median filtering (AMF), followed by the contrast-limited adaptive histogram equalization (CLAHE) process, which removes noise and enhances the image’s contrast. The Swin Transformer is used to identify spatial and spectral features at different levels, while the combination of ConvNeXt-V2 and the squeeze-and-excitation blocks with bidirectional long short-term memory (SE-BiLSTM) classifier refines spatial cues and learns sequential dependencies for precise disease classification. Moreover, the framework includes a Deep Q-Learning (DQL) recommendation module that supplies contextually aware crop management activities, like the optimized use of pesticides and scheduling of irrigation. The experimental results, derived from the SAR-CLD-2024 dataset, indicate a remarkable performance with an accuracy of 99.30%, thereby surpassing existing convolutional neural network (CNN)-based and hybrid models in terms of precision, recall, and F1-score. The combination of deep learning with reinforcement learning in Unmanned Aerial Vehicle (UAV)-based systems indicates a future of sustainable, adaptable, and data-driven precision agriculture.