Activity Preserving Video Face Anonymization

This repository contains the code for the paper Learning to Anonymize Faces for Privacy Preserving Action Detection presented and demonstrated at ECCV 2018.

Zhongzheng Ren, Yong Jae Lee, Michael S. Ryoo
"Learning to Anonymize Faces for Privacy Preserving Action Detection"
in ECCV 2018

Please check the following project page for more details: [Project]

For the training code, please conatact Michael Ryoo (mryoo@egovid.com).

Results

How to use it

1. Preparation

First of all, clone the code

git clone https://github.com/blacknwhite5/Activity-Preserving-Video-Face-Anonymization.git

Then, create a folder

cd Activity-Preserving-Video-Face-Anonymization && mkdir pretrained

2. Prerequisites

python 3.6
pytorch 1.0.0 or higher
CUDA 9.0 or higher

3. Dependencies

Install all the python Dependencies using pip:

pip install -r requirements.txt

torch (if you don't have CUDA 9, look for a version that fit you on PyTorch and install it)
torchvision
cython
cffi
opencv-python
scipy
easydict
matplotlib
pyyaml

4. Download pretrained model

Please download bellow models and place models in pretrained/ folder

SSH-face detector: Google Drive
Privacy Preserving : Google Drive

 pretrained
    ├── check_point.zip
    └── model_1.pth

5. Run real-time demo

You can use a webcam in a real-time demo by running

./scripts/run-demo.sh

Reference

License

The code is for non-commercial academic use and it complies with the following license:

egovid_license

rexnxiaobai/Activity-Preserving-Video-Face-Anonymization