This repository presents a Rice Plant Diseases Classifier that utilizes computer vision feature extraction techniques combined with a fully connected Artificial Neural Network (ANN) for classification.
The main objective of this project is to evaluate the effectiveness of different feature extraction methods for rice disease classification and integrate the best-performing model into a simple web-based application.
ANN Classifier Architecture
The resulting feature vectors were classified using the same defined architecture. Three main experimental setups were conducted:
Experiment 1
Feature extraction techniques used:
- Color Features: HSV Histogram
- Shape Features: Hu Moments
- Texture Features: Haralick Features
📊 Result
Experiment 2
Feature extraction techniques used:
- Color Features: HSV Histogram
- Shape Features: Hu Moments
- Texture Features: Haralick Features
- Spatial Texture Features:Gray Level Co-occurrence Matrix (GLCM)
📊 Result
Experiment 3
Feature extraction techniques used:
- ORB (Oriented FAST and Rotated BRIEF)
- Bag of Visual Words (BoVW)
📊 Result
Based on the experimental results, Experiment 1 performance came on top.
Original Dataset Sources:
Kaggle 1
Kaggle 2
UC Irvine
Mendeley Data
Final classes that are chosen: Healthy, Brownspot, Bacterial Leaf Blight & Leaf Blast
The trained model is integrated into a web-based application designed to be simple and user-friendly.
Upload Image
1. Click to upload the leaf of rice plant image
2. Click "Analyze"
3. The result and insights will be displayed
4. Click "Change Image" to upload another image
