Decentralized Pest Detection in Plants with Blockchain Integrated Machine Learning Models

A significant problem in agriculture is the attack caused by pests on plants. The farmer wants to know which pest is affecting the farm. The farmer can stop the spreading of pests around the whole farm. The challenge of pest detection can only be solved by combining the efforts of those working in Blockchain and Machine Learning (ML) and then storing the resulting pest-detected model alongside the authorized dataset. This paper combines a Deep Learning(DL)-based convolutional neural network (CNN) model with an ML-based XGBoost model trained on the same data to detect pests. Additionally, the data verification is done through Blockchain-based smart contracts which are secure and stored in a decentralized platform. The result shows that DenseNet201 with XGBoost performed better with 96.17% accuracy. This will help the farmers and agricultural experts to find any pest on a farm in a very short time.