yuezunli / DSP-FWA

DSP-FWA: Dual Spatial Pyramid for Exposing Face Warp Artifacts in DeepFake Videos
Other
67 stars 16 forks source link

DSP-FWA: Dual Spatial Pyramid for Exposing Face Warp Artifacts in DeepFake Videos

Overview

We propose an improved version of our previous work "Exposing DeepFake Videos By Detecting Face Warping Artifacts". We employ a dual spatial pyramid strategy on both image and feature level to tackle multi-scale issues.

eval

Requirements

Toy

Check demo.py in toy folder. This script will return a list that contains the "real" probability of each input data. Note we only suppert ResNet-50 based SSPNet model in this version. The checkpoint can be downloaded here.here.(code:x416)

    python demo.py \
    --arch=sppnet \
    --layers=50 \
    --save_dir=../ckpt/ \
    --input=/path/of/video_or_image \
    --ckpt_name=SPP-res50.pth

Citation

Please cite this paper in your publications if this repository helps your research:

@inproceedings{li2019exposing,
  title={Exposing DeepFake Videos By Detecting Face Warping Artifacts},
  author={Li, Yuezun and Lyu, Siwei},
  booktitle={IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)},
  year={2019}
}

Notice

This repository is NOT for commecial use. It is provided "as it is" and we are not responsible for any subsequence of using this code.