Skip to content

ShihaoZhaoZSH/Video-Backdoor-Attack

master
Switch branches/tags

Name already in use

A tag already exists with the provided branch name. Many Git commands accept both tag and branch names, so creating this branch may cause unexpected behavior. Are you sure you want to create this branch?
Code

Latest commit

 

Git stats

Files

Permalink
Failed to load latest commit information.
Type
Name
Latest commit message
Commit time
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Clean-Label Backdoor Attacks on Video Recognition Models

Introduction:

  • Environment: Python3.6.5, TensorFlow-gpu1.14
  • Dataset: UCF101, Model: I3D

Usage:

This is our paper link. You can firstly run train_clean_model.py to get a clean-trained I3D model. The generate_trigger.py and enhance_trigger.py correspond to Backdoor Trigger Generation and Enhancing Backdoor Trigger sections in the paper, respectively. Then you can run train_bad_model.py to train a bad model and run test.py to test it. It's developped based on LossNAN/I3D-Tensorflow.

About

Clean-Label Backdoor Attacks on Video Recognition Models, CVPR2020

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages