RICE LEAF DISEASE DETECTION USING EFFICIENT NET B5 MODEL

Abstract

Plant diseases, particularly those impacting rice leaves, pose a critical challenge for farmers, impeding their ability to meet the escalating food demands of a growing population. The detrimental effects of rice diseases result in significant production and economic losses, profoundly affecting farmers' livelihoods. This distressing situation has unfortunately contributed to an alarming rise in farmer suicides, emphasizing the pressing need for targeted control strategies. To tackle this issue, a novel approach is proposed, leveraging an EfficientNet B5 model, a state-of-the-art convolutional neural network (CNN), for the detection of three major rice diseases: bacterial blight, leaf smut, and brown spot disease. The methodology entails preprocessing and feature extraction techniques, including the utilization of Histogram of Oriented Gradients (HOG) for extracting pertinent features. By harnessing the power of the EfficientNet B5 model, accurate identification of rice diseases can be achieved, empowering farmers to implement effective control measures. Additionally, the analysis of soil content elements is recommended as an ancillary step to gain insights into nutrient deficiencies or imbalances associated with disease occurrence. This comprehensive approach holds great promise in enhancing disease management and enabling farmers to optimize crop health. However, rigorous validation and collaboration with agricultural experts are essential to refine and validate this innovative approach under diverse farming conditions. The integration of advanced technological tools, such as the EfficientNet B5 model, has the potential to mitigate production losses, improve food security, and uplift the well- being of farming communities.