Ensuring Privacy Preservation for Various Plants Multi-product Disease Detection and Pesticides Recommendation Data Using Inception V3

Ensuring the privacy and integrity of detected disease images has become a critical concern due to the increasing reliance on deep-learning algorithms for plant disease detection. Existing vulnerabilities in algorithms to content manipulation raise significant risks of inaccurate disease identification, potentially leading to negative impacts on crop health and economics. Moreover, prevailing models often have limited applicability to specific crops, curtailing their use across diverse agricultural contexts. To tackle these issues, this study presents an innovative approach that integrates deep learning methods with the robust secure hash algorithm (SHA)-256 cryptographic algorithm to safeguard disease-detected image privacy. The proposed model is trained on extensive datasets comprising PlantVillage and Fruits&Vegetables, encompassing a wide range of plants, fruits, vegetables, and leaves from the Krishna district. It achieves an impressive 98% accuracy in detecting diseases across diverse plant types using an Inception V3 convolutional neural network architecture. The model gives a unique hash value to each disease-detected image using the SHA-256 method, assuring privacy and preventing unauthorised access or modification. Additionally, the model’s versatility allows it to identify diseases in a wide range of crop categories, including vegetables, fruits, and their corresponding leaves.The study’s novelty lies in its comprehensive approach, merging advanced deep learning techniques with the robust SHA-256 cryptographic algorithm to ensure precise disease detection and data protection. Furthermore, the model provides pesticide recommendations based on identified diseases, thereby decreasing cyber risks in agriculture, protecting crop health, and reducing economic losses caused by erroneous disease detection and pesticide recommendations.