Convolutional neural network Autoencoder. Implementation based on Cheng et al. Energy Compaction-Based Image Compression Using Convolutional AutoEncoder, Transactions on Multimedia 14 (8), 2019.
Clone this repository and use the pytorch_lightnening container. The extra required packages can be installed using pip from requirements.txt.
Training has been tested on MNIST, ImageNet, and a local histology dataset.
There are a set of configuration files that can be used to set up the training parameters.
Those can be used by passing the argumen -c config.json, where config.json is a json file containing the parameters of the experiment.
The training paramaters can be reviewed by using the following command.
python ./src/train_cae.py -h
The trained model can be tested using the modules src/compress.py and src/decompress.py by separate.
Compression and decompression requires a pre-trained model. Other arguments required to run these modules can be listed with the fllowing commands.
python ./src/compress.py -h
python ./src/decompress.py -h