This repository contains code for hyperspectral image (HSI) classification using a 3D Convolutional Neural Network (CNN) implemented with the SS-DARTS (Single-Stage Differentiable Architecture Search) algorithm. The SS-DARTS algorithm is used to automatically search for an optimal architecture for HSI classification.
- SS-DARTS: Contains the implementation of the SS-DARTS algorithm for architecture search.
- images: Contains images used in the repository, such as diagrams, plots, or visualizations.
- preprocessing: Includes scripts or notebooks for preprocessing HSI data, such as data cleaning, normalization, or dimensionality reduction.
- reference_papers: Contains relevant research papers or articles related to HSI classification, 3D CNNs, and architecture search.
- results: Stores the results, evaluation metrics, or performance analysis obtained from the experiments.
- 3D_CNN_HSI_classification.ipynb: Jupyter Notebook with the implementation of the 3D CNN for HSI classification.
- SS-DARTS.ipynb: Jupyter Notebook demonstrating the implementation of the SS-DARTS algorithm for architecture search.
- Clone the repository:
git clone https://github.com/your-username/HSI-classification-with-3D-CNN-using-SS-DARTS.git
- Set up the necessary dependencies and environment.
- Preprocess the HSI data using the scripts or notebooks in the preprocessing folder.
- Run the .py notebooks in SS-DARTS to train and evaluate the SS-DARTS model for HSI classification using
!python train_HSI.py
- To evaluate the architecture obtained, replace the searched genotype in genotype.py test the model
!python test_HSI.py
This repo contains implementations for four Hyperspectral Image Datasets namely:
This project is licensed under the MIT License.