Exploring traditional and modern techniques in fruit disease detection and classification with IoT integration: A comprehensive survey

Farm produce suffers greatly from fruit diseases, and scaling up the interrogation of crops is not a practical step because of manual identification. Recent trends have emerged regarding various studies on some of the recently discovered computational methodologies used for identification and classification purposes pertaining to fruit disease via application of technologies from machine learning and deep learning. In the beginning, simpler methods, such as SVM and ANN, were successful at their respective tasks but faced problems with feature extraction and generalization. However, the overall accuracy of these models increased with the introduction of newer techniques like CNNs, achieving up to 98.7% in the real-time detection model of strawberry fungal disease. This survey also indicates the possible extent of DL techniques implemented while treating issues related to the diversity of datasets and scalability of models that are necessary for further developing these technologies in agriculture. This survey provides scholars and researchers with wide-ranging information and insights into both traditional and state-of-the-art approaches to fruit disease diagnosis, this survey article can be treated as a treasured reservoir in research academia working in the domain. It provides a comprehensive framework for future research and innovation in agricultural disease control by synthesizing the existing approaches and identifying significant improvements.