This project aims to address significant challenges related to food security and sustainable development by focusing on the early and effective detection of diseases in plant products such as crops and fruits. As we face issues like responsible use of natural resources and climate change, identifying and promptly responding to diseases in agricultural crops becomes essential for promoting sustainable agricultural practices and preserving biodiversity.
The main objective of this project is to develop machine learning models capable of detecting diseases in plant products with high accuracy. This not only promotes the efficiency of agricultural practices but also contributes to reducing the use of chemicals and promoting more sustainable food systems.
Use the Tkinter-based graphical interface application App.py. This application allows you to load the trained models and test images for classification.
Execute the provided Jupyter notebooks to train the models. These include Cnn_base-gridsearch.ipynb, Cnn_base-pso-hyperparameters.ipynb, Cnn_base-random-search.ipynb, and Cnn_train_best_model.ipynb.
The project employs rigorous methodologies including:
Data collection and preparation, experimentation with CNN architectures like VGG16 and ResNet50, Hyperparameter optimization using PSO, Random Search, and Grid Search Training and validation using techniques like Early Stopping and Model Checkpoint.
This work not only significantly improved the model performance in terms of accuracy and generalization but also established a solid foundation for practical applications in real-world environments.