Deep Learning based Automated Rice Plant Disease Recognition and Classification Model

Abstract

Rice plant disease detection and classification include employing methods such as Computer Vision (CV) and Machine Learning (ML) for classifying and recognizing diseases affecting rice crops. These techniques could help agricultural experts and farmers in rapidly managing and diagnosing diseases, eventually resulting in food security and best crop yield. Rice plant disease classification and recognition utilizing Deep Learning (DL) has emerged as an effective strategy to address the problems related to automatic disease detection from the rice plants. DL technique, a subregion of Artificial Intelligence (AI), focuses on training neural networks with different layers to automatically learn intricate representations and patterns from information. This study leverages the Deep Learning-based Automated Rice Plant Disease Recognition and Classification (DL-ARPDRC) approach. The DL-ARPDRC technique encompasses different processes to improve accuracy and diagnosis performance. The primary level encompasses pre-processing stage which includes image resizing and Gaussian Filter (GF) to improve image quality. Next, Otsu’s threshold-based segmentation can be used for the segmentation of the lesion from the background, isolating the region of interest. Feature extraction is accomplished through the utilization of the VGG-19 architecture. Finally, the extracted features are fed into an XGBoost classification model for distinguishing between different types of diseases. A detailed experimental result analysis portrayed that the DL-ARPDRC technique reaches better performance compared to other recent approaches.

Topics & Concepts