Problem: Signatures are used to indicate authenticity but can forged
Solution: use machine learning to detect forgeries
A website at https://uoft-tmi.github.io/signature-forgery-detection/ was created to showcase and interact with the k-nearest neighbours, decision tree, and convolutional neural network models.
Until July 2, 2025, images can be uploaded to the website where the models where be queried. After, the models can be queried locally.
Python 3.10 is required.
- Make a local copy of the GitHub repository
- Open 1 tab in Terminal and in each, navigate to the copy you made in step one
- Then navigate to
website/backend - In that tab, run
python3 manage.py runserver - The endpoint should be available locally and can be run according to our API documentation
This project was made by Puneet Kaur, Steven Lin, Dorothy Lee, Evelyn Phan, Dian Rong, and Angela Zhuo, as a part of the University of Toronto's Truthworthy Machine Intelligence club.