Hybrid Deep Learning Approach for Accurate and Real-Time Detection of Rice and Potato Leaf Diseases

Plant diseases are one of the primary threats to global agricultural production, especially staple crops such as rice and potatoes. Early and accurate disease detection is necessary to reduce losses incurred from crop damage and to ensure food security. This paper presents a deep learning-based leaf disease detection for rice and potato crops using a hybrid model that combines CNNs with ResNet, EfficientNetB3. Such a proposed system utilizes CNNs and ResNet to extract images containing diseased and healthy leaves, with EfficientNetB3 further enhancing the features in feature extraction to provide effective high-level feature representations. Lastly, the Decision Tree is applied to classify features according to relevant disease categories, ensuring interpretability and reliability in real-time disease detection. The model is trained and tested on a comprehensive dataset covering the diseases of Rice Blast, Brown Spot, Sheath Blight for rice, and Late Blight, Early Blight, and Powdery Scab on potatoes. Results have shown that this hybrid approach would achieve high accuracy of 98.5%. This study provides real-time disease classification and actionable recommendations for farmers. Its focus will be on better control of diseases, lowered dependency on traditional practices, and optimization of crop yields coupled with further precision agriculture with sustainability.