Exploration of rice disease data sharing scheme based on blockchain and federal learning

Rice diseases severely impact rice yield and quality, posing threats to food security and agricultural economies. Traditional manual identification of rice diseases exhibits multiple limitations, while deep learning, despite its potential, faces challenges in agricultural applications such as insufficient data volume, uneven data distribution, and data privacy-security concerns. This study proposes a rice disease data sharing and privacy protection scheme by integrating blockchain technology with a federated learning framework. Leveraging blockchain’s decentralized and tamper-proof features to replace traditional federated learning’s centralized servers, the approach employs smart contracts for global model aggregation. A differential privacy mechanism is introduced to process model parameters, ensuring data privacy protection. Experimental results demonstrate that this solution effectively safeguards data privacy without compromising federated learning accuracy, while enhancing data utilization efficiency and sharing effectiveness, thereby providing robust technical support for the advancement of smart agriculture.