Taksir/Homoglyph

This repository contains code for the homoglyph project.

Convolutional Neural Network Based Ensemble Approach for Homoglyph Recognition

1. First download the dataset from here. This dataset contains training and validation set.
2. Go above your current directory
3. Create a folder named "DataSet"
4. Paste the folder named "Custom" (which you just downloaded) inside it.
5. run: "pip install -r requirements.txt" in your shell.
6. go back to the main directory
7. run: python scratch1.py
8. run: python scratch2.py
9. After you have properly trained a good model (>98% accuracy on validation set), download final test dataset from here.
10. Code for testing an image can be found as commented out in line: 325-355 in scratch1.py file.
11. Code for testing an image can be found as commented out in line: 261-390 in scratch2.py file.
12. Modify codes accordingly to evaluate on the final test dataset.
13. For running the xfer_1.py and xfer_2.py files, first split files in the "Custom" folders into "Train" and "Test" folder in same directory. Line: 20-21 in xfer_1.py and line: 26-27 in xfer_2.py should be changed according to the number of files that you have put in the folder.
14. run: python xfer_1.py
15. run: python xfer_2.py
16. Evaluate the obtained model on the previous final test dataset.