This project is a deep learning model that identifies whether a plant is healthy or not by analyzing photos of its leaves. It's trained to classify plants into various categories based on their health status.
The model was trained on the PlantVillage dataset, which includes the following classes:
- Pepper__bell___Bacterial_spot
- Pepper__bell___healthy
- Potato___Early_blight
- Potato___Late_blight
- Potato___healthy
- Tomato_Bacterial_spot
- Tomato_Early_blight
- Tomato_Late_blight
- Tomato_Leaf_Mold
- Tomato_Septoria_leaf_spot
- Tomato_Spider_mites_Two_spotted_spider_mite
- Tomato__Target_Spot
- Tomato__Tomato_YellowLeaf__Curl_Virus
- Tomato__Tomato_mosaic_virus
- Tomato_healthy
https://www.kaggle.com/datasets/arjuntejaswi/plant-village- Clone this repository:
git clone https://github.com/biswadeep_roy/Plant-Health-Classifier.gitInstall the required libraries:
pip install -r tensorflow
pip install matplotlib
Open the Jupyter Notebook (Plant_Health_Classifier.ipynb) to see the code and run the model.
Follow the instructions in the notebook to train or use the pre-trained model.
The model achieved an accuracy of 98.39% on the validation dataset.
/data: Directory for data sources and preprocessing scripts. /notebooks: Contains Jupyter Notebook for model development and evaluation. /models: Model checkpoints and saved models. /scripts: Additional Python scripts if necessary.
To obtain the PlantVillage dataset, visit the data source. In the data directory, you can find preprocessing scripts and instructions to prepare the dataset for training.
If hyperparameter tuning was performed, the best hyperparameters and the reasons for choosing them are documented in the Jupyter Notebook.
You can download pre-trained model checkpoints from the /models directory. Use these checkpoints for inference or further training.
The code includes error handling and informative error messages for common issues users might encounter.
To ensure that the code is working as expected, unit tests have been implemented. You can find them in the /tests directory.
If you use existing research or datasets, provide proper citations in your work to give credit to the original authors and sources.
The Jupyter Notebook contains various visualizations, including sample predictions, training curves, and data analysis to help users understand the project better.
Consider adding a section that compares your model's performance with other existing models or research in the field, providing users with context for your results.
It is a real-world application and can be deployed to GCD, AWS or any other cloud service
If you would like to contribute to this project, please fork this repository, make your changes, and submit a pull request.
This project is licensed under the MIT License.