International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 11 Issue: 04 | Apr 2024
p-ISSN: 2395-0072
www.irjet.net
RICE LEAF DISEASE DETECTION USING DEEP LEARNING B. Jagadeesh1, A. Umesh Chandra2, G. Uday Sankar Reddy3, G. Satish4, Md. Shakeel Ahmed5 1,2,3,4Department of Information Technology, VVIT, AP, India
5Associate Professor, Dept. of Information Technology, VVIT, AP, India
---------------------------------------------------------------------***--------------------------------------------------------------------the potential of deep learning in resolving complex Abstractagricultural challenges.
Rice, a vital staple crop globally, confronts significant threats from various leaf diseases, adversely impacting yield and agricultural sustainability. In response, our project endeavors to develop an efficient rice leaf disease detection system using deep learning models. The study utilizes a comprehensive dataset encompassing images of rice leaves afflicted with four prevalent diseases: bacterial blight, blast, tungro, and brown spot. Leveraging diverse deep learning architectures, including Simple CNN, ResNet, Inception, and LeNet, we train and evaluate models to accurately classify these diseases. Through meticulous experimentation, we identify the model exhibiting the highest accuracy on the validation dataset. Deploying the selected model with the Streamlit library, we construct a userfriendly frontend interface facilitating seamless interaction. This interface empowers users to upload rice leaf images, enabling real-time disease prediction and identification. By integrating advanced deep learning techniques with accessible user interaction, our project offers a practical solution for farmers and agricultural experts to swiftly diagnose and address rice leaf diseases. This initiative holds promise for enhancing crop management practices and bolstering agricultural productivity.
2. EXISTING SYSTEM Existing systems for rice leaf disease detection often rely on text datasets and machine learning algorithms, such as Support Vector Machine (SVM) or K-Nearest Neighbors (KNN). These systems typically suffer from limitations in handling image complexity, potentially leading to suboptimal classification results. Some existing systems may utilize only one algorithm, which can further limit their accuracy and robustness. Despite their contributions to disease detection, these systems are often slower and less efficient compared to more advanced techniques like deep learning. Hence, there is a pressing need to explore and implement more sophisticated approaches to enhance the effectiveness of rice leaf disease detection systems.
Key Words: Deep Learning, Image classification, Simple CNN, LeNet, ResNet, Inception, Bacterial Blight, Blast, Tungro, Brownspot
1.INTRODUCTION Rice leaf diseases pose significant threats to crop yields worldwide, necessitating faster and more reliable detection methods. Current approaches relying on textual datasets and traditional algorithms like SVM/KNN face challenges with image complexity, potentially leading to suboptimal classification. Recent technological advancements in image processing have opened avenues for automated disease detection. Our project aims to enhance the accuracy and efficiency of rice leaf disease prediction by integrating deep learning techniques, particularly Convolutional Neural Networks (CNNs). By directly tackling the pressing issue of rice leaf diseases, our system offers a practical tool for farmers to improve disease management and minimize crop losses. Designed to be user-friendly, our system provides farmers with a convenient means of predicting and managing rice leaf diseases. The integration of CNN models not only addresses rice leaf diseases but also underscores
© 2024, IRJET
|
Impact Factor value: 8.226
Fig: Existing System Architecture
3. PROPOSED SYSTEM The proposed system for rice leaf disease detection leverages the power of deep learning models, including Simple CNN, ResNet, LeNet, and Inception, to enhance classification accuracy. Unlike existing systems, which often rely on text datasets and traditional machine learning algorithms, our approach utilizes image datasets, enabling more effective analysis of the complex visual features present in rice leaf images. Through rigorous experimentation and comparison of model accuracies, we
|
ISO 9001:2008 Certified Journal
|
Page 62