Rice Leaf Disease Detection

This repository contains a solution aimed at improving the accuracy and reliability of bacterial leaf blight detection in rice crops through the use of computer vision (CV) techniques. By implementing image processing operations for preprocessing input images, we ensure optimal feature extraction, enabling the CV model to effectively handle real-world agricultural data.

Implemented CV Techniques

This project implements various computer vision (CV) techniques to enhance the accuracy and reliability of bacterial leaf blight detection in rice crops. The implemented techniques include:

Preprocessing Techniques

Noise Reduction (Gaussian Blur): Utilized to reduce noise and smoothen images.

Space Transformation: Applied for transforming color spaces or resizing images.

Histogram Equalization: Employed to enhance the contrast of images.

Feature Extraction Techniques

Edge Detection: Implemented using Sobel and Canny edge detection algorithms.

Morphological Operations: Utilized dilation and erosion operations along with connected components analysis.

Region-based Segmentation: Employed for segmenting regions of interest in images.

Model Development

Convolutional Neural Networks (CNNs): Developed CNN models for pattern recognition in leaf blight detection.

Data Augmentation: Implemented strategies to augment the dataset, increasing the model's generalizability.

Model Optimization: Techniques such as hyperparameter tuning and regularization were applied to optimize the performance of the models.

Usage

The implemented CV techniques are utilized within the application for bacterial leaf blight detection in rice crops. Users can leverage these techniques by uploading images through the provided interface to receive actionable insights for disease intervention.

PROJECT UNDER DEVELOPMENT