hsid-cnn

Tensorflow 2.0 Implementation of HSID-CNN for denoising hyperspectral images.

• Dataset Preparation & Environment Preparation

  • Prepare two file train.txt & valid.txt with the paths to hyperspectral images (see data folder)
  • Add the paths of the files into a config file, sample_config.yaml is an example.
  • Install the requirements with pip install -r requirements.txt

• Training

  • Hyperparameters for training can be tweaked in the config file or you can use the default ones from sample_config.yaml.
  • Run python train.py --config-file ./sample_config.yaml for running the script
  • For using a pre-trained model: modify the checkpoint_path variable inside your config file

• Summaries

  • For having a look at the various error plots: run tensorboard --logdir='./summaries'

• Original Paper: HSID-CNN

Note -- I cannot release the pretrained models publically, if you want it, please shoot me an email. The code is strictly for academic purposes only!