This project focuses on Land Use and Land Cover (LULC) classification using a shallow CNN. The main objective was to integrate domain knowledge of Remote Sensing to simplify the model architecture and reduce the training duration without significantly compromising the results.
The LULC classification was performed on the Indian Pines dataset, a widely-used hyperspectral dataset in the remote sensing community. By incorporating domain-specific insights, we were able to streamline the model, achieving high accuracy while minimizing the training time.
- Training Accuracy: 97.95%
- Testing Accuracy: 97.45%
These results were achieved by carefully designing the model architecture and selecting features that leverage remote sensing knowledge.
To run this project, simply import the Project.ipynb to Google Colab
- Google Colab
Detailed documentation of the steps involved in the model training and testing process can be found in the docs/ directory. This includes:
- Data preprocessing
- Model architecture design
- Training process
- Performance evaluation
The model was trained and tested on the Indian Pines dataset, achieving a training accuracy of 97.95% and a testing accuracy of 97.45%. The results indicate that the approach taken effectively balances model complexity and accuracy.
This project is licensed under the MIT License.