Implementation of IoT with image processing in plant monitoring system

Abstract

This research demonstrates the importance of image processing and IoT in plant monitoring systems. The study focused on pest identification using Mobilenet_v2 and Resnet_v2_50 models trained with data sets of five pests. The accuracy was 92.12% and 82% respectively, but it can be improved by adding more data sets and using preprocessing techniques. The identification of pests can help farmers take the appropriate actions to increase production. The research also utilized YoloV4 and augmentation methods for semantic and instance segmentation, achieving a 54.76% MAP on the test set, demonstrating the reliability of physical indicators of plant pathogens. The proposed system is efficient and simple, and the accuracy can be improved by using different image processing techniques. The study suggests that image processing techniques and IoT can be crucial in plant monitoring systems, leading to increased productivity and food quality.