Less-supervised learning with knowledge distillation for sperm morphology analysis

This project is the official implementation of the paper "Less-supervised learning with knowledge distillation for sperm morphology analysis".

Configuration

You can modify the hyperparameters and other settings in the config.py file.

Project Structure

• data/: Directory for storing the MHSMA dataset
• models/: Contains the VGG and custom VGG model implementations
• utils/: Utility functions for data loading, loss calculation, and attacks
• train.py: Script for training the model
• test.py: Script for testing the model
• config.py: Configuration file with hyperparameters and settings

Data Directory

Place the MHSMA dataset files in this directory. The expected files are:

• x_64_train.npy
• x_64_valid.npy
• x_64_test.npy
• y_acrosome_train.npy
• y_acrosome_valid.npy
• y_acrosome_test.npy

Make sure to download these files from the official MHSMA dataset source and place them in this directory before running the training or testing scripts.

Citation

If you use this code in your research, please cite the following paper:

@article{doi:10.1080/21681163.2024.2347978,
        author = {Ali Nabipour, Mohammad Javad Shams Nejati, Yasaman Boreshban and Seyed Abolghasem Mirroshandel},
        title = {Less-supervised learning with knowledge distillation for sperm morphology analysis},
        journal = {Computer Methods in Biomechanics and Biomedical Engineering: Imaging \& Visualization},
        volume = {12},
        number = {1},
        pages = {2347978},
        year = {2024},
        publisher = {Taylor \& Francis},
        doi = {10.1080/21681163.2024.2347978},
        URL = {https://doi.org/10.1080/21681163.2024.2347978}
}