cassidylaidlaw/perceptual-advex

Code and data for the ICLR 2021 paper "Perceptual Adversarial Robustness: Defense Against Unseen Threat Models".

48
/ 100
Emerging

This project helps machine learning researchers and practitioners understand and defend against 'perceptual adversarial attacks' on image classification models. You input an image classification model and a dataset, and it outputs images that have been subtly manipulated to fool the model, along with tools to train models that are more robust to these kinds of attacks. It's for anyone building or evaluating image recognition systems who needs to ensure their models are secure against sophisticated visual trickery.

No commits in the last 6 months. Available on PyPI.

Use this if you are developing or testing image classification models and need to assess their vulnerability to visually imperceptible, but model-disrupting, alterations in images.

Not ideal if you are looking for general-purpose image augmentation or data preprocessing tools unrelated to adversarial robustness.

computer-vision image-classification machine-learning-security adversarial-machine-learning
Stale 6m
Maintenance 0 / 25
Adoption 8 / 25
Maturity 25 / 25
Community 15 / 25

How are scores calculated?

Stars

56

Forks

9

Language

Python

License

MIT

Last pushed

Jan 18, 2022

Commits (30d)

0

Dependencies

8

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/cassidylaidlaw/perceptual-advex"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.