Skip to content

Latest commit

 

History

History
60 lines (45 loc) · 3.48 KB

README.md

File metadata and controls

60 lines (45 loc) · 3.48 KB

Xception Network trained with CloGAN for unsupervised domain adaptation

python 3 tensorflow 2

The project directory consists of files, programmed in Python. The description of each file is commented in the beginning of the corresponding file.

It s based on my thesis that is uploaded to Academia.edu. The main proposals introduced here are CloGAN which is GAN-based multi-label domain adaptation approach and Robust Feature Mapping metrics to analyze the influence of domain adaptation approaches to the feature mapping of neural network.

Table of Contents

Why?

Supervised labeling of the training data is often scarce in a lot of task. The medical classification task is no exception. The main reason to develop this technique of unsupervised domain adaptation is to utilize the unlabelled training data to strengthen the network in classifying diseases. This also encourages the network to learn over other similar domains, e.g. X-ray images from other hospitals.

Testing

Before testing, make sure that the weights of the Xception network is prepared. To have the pre-trained model, download https://github.com/samuelmat19/CloGAN/releases/download/pretrained-model/model.hdf5 and copy it to "resources" directory.

Predict abnormalities/symptoms in chest X-ray image with Xception network, trained on classification loss and CloGAN

  1. Make sure that Python 3 is installed in your system.

  2. Open terminal or command line that has access to the Python binary and change the directory to this project.

  3. Install Python virtual environment : pip3 install --user virtualenv, in case you do not have the virtual environment installed.

  4. Create new virtual environment : python3 -m venv CloGAN

  5. Activate the virtual environment : In Linux and macOS, run source CloGAN/bin/activate and in Windows, .\CloGAN\Scripts\activate

  6. Install the required libraries in the virtual environment : pip install -r requirements.txt. Note that the tensorflow in the requirements.txt is CPU version.

  7. Run the CloGAN_predict.py with : python CloGAN_predict.py

  8. Input the file path of the chest X-ray image to be predicted. (There are two examples that you could use, taken from ChestXray-14: resources/cardiomegaly.png and resources/effusion.png)

  9. The predictions with saliency maps will be generated.

  10. Enter exit if you want to leave the app.

  11. You can leave the virtual environment : deactivate

CONTRIBUTING

To contribute to the project, these steps can be followed. Anyone that contributes will surely be recognized and mentioned here!

Contributions to the project are made using the "Fork & Pull" model. The typical steps would be:

  1. create an account on github
  2. fork this repository
  3. make a local clone
  4. make changes on the local copy
  5. commit changes git commit -m "my message"
  6. push to your GitHub account: git push origin
  7. create a Pull Request (PR) from your GitHub fork (go to your fork's webpage and click on "Pull Request." You can then add a message to describe your proposal.)

LICENSE

This open-source project is licensed under MIT License.