Smart Irrigation and Pest Monitoring System Combining IoT, Ethereum Smart Contracts and ResNeSt-DDETR

Abstract

The pests and the ideal irrigation should be monitored simultaneously so that the crops can be efficiently managed to yield the maximum. The intended dual-purpose solution to the pest detection problem, which is proposed in this study, is the combination of IoT-enabled sensors with Ethereum smart contracts and a deep learning-based ResNeSt-DDETR pest detector. The ResNeSt backbone is able to improve the extraction of features with the help of split-attention mechanisms, whereas Deformable DETR pays attention to the areas which are of interest in order to achieve precise detection in the field under complex conditions. IoT sensors constantly check soil moisture, temperature, and humidity to adjust the irrigation patterns to control the water management accurately. The pest detections and irrigation logs are registered safely on the Ethereum blockchain and provide a solution with tamper-proof, transparent, and traceable data. The system is deployed on edge devices and implemented on Python with PyTorch, OpenCV, and Web3.py and works in real time. The experimental assessment of the IP102 data reveals that the model has a high accuracy (95.2%), precision (94.5%), recall (93.7%), F1-score (94.1%), and mAP 0.5:0.95 = 90.6% indicating that it is effective in integrated pest and irrigation management in precision agriculture.