This project, titled "Pixel-Wise Classification of Multispectral Satellite Images Using Machine Learning", presents a cloud-based application for pixel-wise classification of multispectral satellite images using machine learning. It supports both supervised (KNN with Euclidean distance) and unsupervised (K-means) classification, enabling users to upload GeoTIFF images, select training samples, and classify land cover types. The application features a user-friendly interface designed to simplified the process of image classification.
- GeoTIFF Image Upload: Easily upload multispectral GeoTIFF images for analysis.
- Training Sample Selection: Interactively select training samples for supervised classification.
- Supervised Classification: Utilize K-Nearest Neighbors algorithm with Euclidean distance for accurate land cover classification.
- Unsupervised Classification: Apply K-means clustering to identify natural groupings in the data.
- Visualization: View and analyze classification results directly within the application.
The application has been published via Microsoft Azure. You can click the link to access the application directly. However, because it uses a free server, the performance of the web version is not very good. For a better experience, it is recommended to use the following method.
-
Clone the Repository:
git clone https://github.com/zyl009/Pixel-Wise-Classification-of-Multispectral-Satellite-Images-Using-Machine-Learning.git cd Pixel-Wise-Classification-of-Multispectral-Satellite-Images-Using-Machine-Learning -
Set Up a Virtual Environment:
python3 -m venv env source env/bin/activate # On Windows, use `env\Scripts\activate`
-
Install Dependencies:
pip install -r requirements.txt
-
Run the Application:
python app.py
-
Access the Web Interface: Open your web browser and navigate to
http://localhost:5000. -
Upload and Classify Images:
- Upload your multispectral GeoTIFF image.
- Choose the desired classification method (KNN or K-means).
- For supervised classification, select training samples on the image. For unsupervised classification, input the number of classes.
- View and analyze the resulting classified image.
See the report and slides file for more details.
This project utilizes the following open-source libraries:
| Library | Description | License Type |
|---|---|---|
| Flask | Micro web framework for Python | BSD License |
| NumPy | Library for numerical computations | BSD License |
| Rasterio | Access and process raster data (e.g., GeoTIFF) | BSD License |
| Pillow (PIL) | Image processing library | HPND License |
| Matplotlib | Plotting and visualization library | PSF/BSD License |
| Shapely | Geometric objects and operations | BSD License |
| Scikit-learn | Machine learning library | BSD License |
| Base64 (built-in) | Encoding and decoding base64 strings | Python Standard |
| io (built-in) | Core input/output functionality | Python Standard |
- Pillow: Copyright © 2010-2024 by Alex Clark and contributors. Licensed under the HPND License.
Each library is subject to its respective license.Refer to their documentation for more details.
Contributions are welcome! Please fork the repository and create a pull request with your changes. Ensure that your code adheres to the project's coding standards and includes appropriate tests.
This project is licensed under the MIT License. See the LICENSE file for more details.
- Liang Zhongyou – GitHub Profile