Iot and Machine Learning-Based Sugarcane Leaf Disease Classification

Leaf infections have a significant impact on sugarcane production, a valuable cash crop, and can result in substantial financial losses. Traditional methods of identifying illnesses are time-consuming and need specific expertise. In this research, we propose a machine learning (ML) technique based on quantum-behavior particle swarm optimization (QPSO) and image processing techniques for accurate disease detection, aiming to provide IoT-integrated sugarcane leaf disease prediction. Furthermore, the IoT sensors are connected to the camera modules to collect environmental data (temperature, humidity, and soil moisture) and gather images of sugarcane leaves. A Contrast-Limited Adaptive Histogram Equalization (CLAHE) algorithm is used in pre-processing to enhance contrast. After that, leaf segmentation is performed using a watershed algorithm to isolate the affected areas. The use of a lightweight deep learning model, DenseNet, for feature extraction is optimized for edge computing. Furthermore, the optimal hyperparameter selection can be done by classifying using a convolutional neural network (CNN) and a support vector machine (SVM), with improved performance through QPSO based on extracted features. Additionally, the implemented results are sent to a cloud-based or edge computing platform, allowing farmers to access disease predictions through a mobile or web-based dashboard. Furthermore, it provides alerts and preventive measures to reduce crop losses. This system offers a cost-effective, scalable, and efficient solution for precision agriculture.