Plant diseases are a common issue in agriculture that can have significant negative effects on crop quality and productivity. This project focuses on the automatic detection of plant diseases using machine learning and image processing techniques. Early disease detection is crucial to prevent large-scale outbreaks and minimize the impact on crops.
The scope of this project is to develop a machine learning and image processing-based application that can detect diseases in plant leaves uploaded by users. The application aims to assist farmers and researchers in quickly identifying leaf diseases, which can lead to timely actions and the production of bio products based on the identified diseases.
The main objective of this project is to analyze and identify leaf diseases using image processing and machine learning. The project aims to help farmers and researchers by providing accurate disease detection, which can improve the quality and productivity of agricultural products.
- Python
- OpenCV library
- Scikit learn library (SVM)
- SKlearn
- Numpy
- Mathlib Library
Machine Learning
The implementation strategy includes several steps:
- RGB image acquisition
- Convert RGB to HSI format
- Mask green pixels
- Segment components
- Extract useful segments
- Evaluate feature parameters
- Configure SVM for disease detection
Step 1: Run GUIdriver.py for image segmentation and feature extraction.
Step 2: Call classifier.py in main.py for classifying the leaf in the input image as "infected" or "healthy".
To create the dataset, run leafdetectionALLsametype.py for images of the same category (infected/healthy) and leafdetectionALLmix.py for a mixed dataset of both categories. The code runs on .jpg, .jpeg, and .png image formats in the specified directory.
Support Vector Machine (SVM) is used as the classification model. Tuning parameters include Kernel, Regularization, Gamma, and Margin. SVM provides accurate classification by creating an optimal hyperplane that separates different classes.
- Low-resolution photos or improper lighting may lead to incorrect results.
- Certain leaf conditions, like a broken midrib, might result in false positives.
- Implementation on Raspberry Pi with PiCamera for live input.
- Exploring alternate methods to improve disease identification accuracy.
📂 Explore Project Artifacts: If you're curious about our project's journey, including presentation slides and comprehensive reports, take a stroll through our Google Drive repository.
🔍 A Closer Look at Operation: Delve deeper into the workings of our system! The report provides an immersive experience, guiding you through each step. You'll find an assortment of vibrant screenshots alongside step-by-step instructions, allowing you to witness the system's functionality firsthand.
Enjoy the journey of discovery and exploration! 🚀