rohit-kundu/two-step-feature-enhancement

Based on our paper "Cervical Cytology Classification Using PCA & GWO Enhanced Deep Features Selection" published in SN Computer Science

Cervical Cytology Classification Using PCA & GWO Enhanced Deep Features Selection

Official Python Implementation of the paper titled "Cervical Cytology Classification Using PCA & GWO Enhanced Deep Features Selection" published in the special issue "AI and Deep Learning Trends in Healthcare" of SN Computer Science.

Requirements

To install the required dependencies run the following using the Command Prompt:

pip install -r requirements.txt

Implementing the code for Cervical Cytology data

1. Herlev Pap Smear dataset by Jantzen et al.: http://mde-lab.aegean.gr/index.php/downloads
2. Mendeley Liquid Based Cytology dataset by Hussain et al.: https://data.mendeley.com/datasets/zddtpgzv63/4
3. SIPaKMeD Pap Smear dataset by Plissiti et al.: https://www.cs.uoi.gr/~marina/sipakmed.html

Structure the directory as follows:

.
+-- data
|   +-- .
|   +-- train
|   +-- val
+-- extract_features.py
+-- fitnessFUNs.py
+-- GWO.py
+-- main.py
+-- resnet50.csv
+-- selector.py
+-- solution.py
+-- transfer_functions_benchmark.py

To extract ResNet-50 features run the following script:

python extract_features.py

Similarly the script can be modified for extracting features from other models.

Run the following code for the feature set optimization:

python main.py --num_csv 2

Set num_csv to the number of features csv files you have. You will be asked to enter the names of the csv files upon executing the above code. Execute python main.py -h to get the details of all the available arguments.

Citation

If this repository helps you in your research in any way, please cite our paper:

@article{Basak2021,
      author={Basak, Hritam and Kundu, Rohit and Chakraborty, Sukanta and Das, Nibaran},
      title={Cervical Cytology Classification Using PCA and GWO Enhanced Deep Features Selection},
      journal={SN Computer Science},
      year={2021},
      month={Jul},
      day={07},
      volume={2},
      number={5},
      pages={369},
      issn={2661-8907},
      doi={10.1007/s42979-021-00741-2},
      url={https://doi.org/10.1007/s42979-021-00741-2}
}

two-step-feature-enhancement's Issues

There are a lot of bugs in the code

I hope the author can improve it and specify the data set format of the code. Thanks @Rohit-Kundu

error

The project has an error and is not executable
parser.add_argument('--num_csv', type=int, required = True, help='How many csv files of features do you have?')

code optimization suggestion

Some code optimization suggestions in extract_features.py