AI-Powered Crop Monitoring and Management with Blockchain Traceability
Pages:38-49
Shivam Suryawanshi, Abhijit Padghan, Sujal Nighot, Tejal Pawar, Prof. Nalini Tiwari
This paper presents a simple and compact system for the early detection of diseases in soybean crops. The system allows farmers to take a photo of a soybean leaf and upload it through a user-friendly web interface. The uploaded image is then processed by the backend, which analyzes its color and texture features to identify the condition of the leaf. Based on this analysis, the system classifies the leaf into one of three categories: healthy, affected by Bacterial Blight, or damaged by Diabrotica speciosa. After the leaf is classified, the result, along with its confidence score and the time of detection, is securely stored in a blockchain ledger. Each record is connected to the previous one using SHA-256 hashing, making the ledger tamper-resistant. If any stored record is modified, the chain is broken, making the change easy to detect. A React.js dashboard displays the classification result immediately after processing and allows users to verify that the stored records remain unchanged. To evaluate the system, a dataset of 800 labelled soybean leaf images was used. Out of these, 640 images were used to train the model, while the remaining 160 images were reserved for testing its performance. On the test dataset, the model achieved an accuracy of 89.66% and a weighted F1-score of 0.88, demonstrating reliable performance in classifying soybean leaf conditions. The system was also efficient, with each prediction taking less than 100 milliseconds on standard consumer hardware. To evaluate the security of the blockchain, 100 consecutive transactions were recorded, and all were stored successfully without any integrity issues. By combining AI-based disease detection with blockchain technology, the system provides farmers with more than just a diagnosis. Each detection result is securely recorded in a tamper-resistant ledger, ensuring that any attempt to modify the stored information can be identified immediately. This makes the system both reliable for disease detection and trustworthy for maintaining secure records.
Support Vector Machine, Blockchain Traceability, Crop Disease Detection, Proof-of-Work, Precision Agriculture, Leaf Image Classification, Flask, React.js.