Implementation for the IJCAI2021 work "Beyond the Spectrum: Detecting Deepfakes via Re-synthesis"

Overview

Beyond the Spectrum

Implementation for the IJCAI2021 work "Beyond the Spectrum: Detecting Deepfakes via Re-synthesis" by Yang He, Ning Yu, Margret Keuper and Mario Fritz.

Pretrained Models

We release the model trained on CelebA-HQ dataset with image resolution 1024x1024. For the super resolution, we use 25,000 real images from CelebA-HQ to train it. For the detectors, we use 25,000 real images and 25,000 fake images to train a binary classifier based on ResNet-50.

We release some models as examples to show how to apply our models based on pixel-level or stage5-level reconstruction errors to detect deepfakes. Download link: https://drive.google.com/file/d/1FeIgABjBpjtnXT-Hl6p5a5lpZxINzXwv/view?usp=sharing.

If you have further questions regarding the trained models, please feel free to contact.

Train

  1. Train the super resolution model.

We use Residual Dense Network (RDN) in our work. The following script shows the hyperparameters used in our experiments. To be noticed, we only use 4 images to show how to run the script. For simplicity, you can download the pretrained model from the above link.

bash script/train_super_resolution_celeba.sh [GPU_ID]
  1. Train the detectors.

After obtaining the super resolution, we use pixel-level or stage5-level L1 based recontruction error to train a classifier. The following scripts use 10 training example to show how to train a classifier with a given super resolution model. You may need to adjust the learning rate and number of training epochs in your case.

bash script/train_pixel_pggan.sh [GPU_ID]
  1. Finetune with auxiliary tasks

In order to improve the robustness of our detectors, we introduce auxiliary tasks (i.e., colorization or denoising) into the super resolution model training and finetune the whole model end-to-end. The following scripts show how to train a model with those tasks.

bash script/train_pixel_pggan_colorization.sh [GPU_ID]
bash script/train_stage5_stylegan_denoising.sh [GPU_ID]

Test

Please download our models. You can use pixel-level or stage5-level to perform deepfakes detection.

bash script/test_pixel_celeba.sh [GPU_ID]
bash script/test_stage5_celeba.sh [GPU_ID]

Citation

If our work is useful for you, please cite our paper:

@inproceedings{yang_ijcai21,
  title={Beyond the Spectrum: Detecting Deepfakes via Re-synthesis},
  author={Yang He and Ning Yu and Margret Keuper and Mario Fritz},
  booktitle={30th International Joint Conference on Artificial Intelligence (IJCAI)},
  year={2021}
}

Contact: Yang He ([email protected])

Last update: 08-22-2021

Owner
Yang He
Applied Scientist in Amazon Last Mile PostDoc in CISPA Helmholtz Center for Information Security / PhD in Max Planck Institute for Informatics
Yang He
A curated list of resources for Image and Video Deblurring

A curated list of resources for Image and Video Deblurring

Subeesh Vasu 1.7k Jan 01, 2023
Wordle-solver - Wordle answer generation program in python

🟨 Wordle Solver 🟩 Wordle answer generation program in python ✔️ Requirements U

Dahyun Kang 4 May 28, 2022
3rd Place Solution for ICCV 2021 Workshop SSLAD Track 3A - Continual Learning Classification Challenge

Online Continual Learning via Multiple Deep Metric Learning and Uncertainty-guided Episodic Memory Replay 3rd Place Solution for ICCV 2021 Workshop SS

Rifki Kurniawan 6 Nov 10, 2022
Python scripts form performing stereo depth estimation using the CoEx model in ONNX.

ONNX-CoEx-Stereo-Depth-estimation Python scripts form performing stereo depth estimation using the CoEx model in ONNX. Stereo depth estimation on the

Ibai Gorordo 8 Dec 29, 2022
This repository contains a PyTorch implementation of "AD-NeRF: Audio Driven Neural Radiance Fields for Talking Head Synthesis".

AD-NeRF: Audio Driven Neural Radiance Fields for Talking Head Synthesis | Project Page | Paper | PyTorch implementation for the paper "AD-NeRF: Audio

551 Dec 29, 2022
Object detection GUI based on PaddleDetection

PP-Tracking GUI界面测试版 本项目是基于飞桨开源的实时跟踪系统PP-Tracking开发的可视化界面 在PaddlePaddle中加入pyqt进行GUI页面研发,可使得整个训练过程可视化,并通过GUI界面进行调参,模型预测,视频输出等,通过多种类型的识别,简化整体预测流程。 GUI界面

杨毓栋 68 Jan 02, 2023
Official code for 'Pixel-wise Energy-biased Abstention Learning for Anomaly Segmentationon Complex Urban Driving Scenes'

PEBAL This repo contains the Pytorch implementation of our paper: Pixel-wise Energy-biased Abstention Learning for Anomaly Segmentationon Complex Urba

Yu Tian 115 Dec 29, 2022
Official Keras Implementation for UNet++ in IEEE Transactions on Medical Imaging and DLMIA 2018

UNet++: A Nested U-Net Architecture for Medical Image Segmentation UNet++ is a new general purpose image segmentation architecture for more accurate i

Zongwei Zhou 1.8k Jan 07, 2023
A MatConvNet-based implementation of the Fully-Convolutional Networks for image segmentation

MatConvNet implementation of the FCN models for semantic segmentation This package contains an implementation of the FCN models (training and evaluati

VLFeat.org 175 Feb 18, 2022
Permute Me Softly: Learning Soft Permutations for Graph Representations

Permute Me Softly: Learning Soft Permutations for Graph Representations

Giannis Nikolentzos 7 Jul 10, 2022
ConvMAE: Masked Convolution Meets Masked Autoencoders

ConvMAE ConvMAE: Masked Convolution Meets Masked Autoencoders Peng Gao1, Teli Ma1, Hongsheng Li2, Jifeng Dai3, Yu Qiao1, 1 Shanghai AI Laboratory, 2 M

Alpha VL Team of Shanghai AI Lab 345 Jan 08, 2023
This is the implementation of GGHL (A General Gaussian Heatmap Labeling for Arbitrary-Oriented Object Detection)

GGHL: A General Gaussian Heatmap Labeling for Arbitrary-Oriented Object Detection This is the implementation of GGHL 👋 👋 👋 [Arxiv] [Google Drive][B

551 Dec 31, 2022
the official implementation of the paper "Isometric Multi-Shape Matching" (CVPR 2021)

Isometric Multi-Shape Matching (IsoMuSh) Paper-CVF | Paper-arXiv | Video | Code Citation If you find our work useful in your research, please consider

Maolin Gao 9 Jul 17, 2022
A PoC Corporation Relationship Knowledge Graph System on top of Nebula Graph.

Corp-Rel is a PoC of Corpartion Relationship Knowledge Graph System. It's built on top of the Open Source Graph Database: Nebula Graph with a dataset

Wey Gu 20 Dec 11, 2022
Code repository for the paper "Doubly-Trained Adversarial Data Augmentation for Neural Machine Translation" with instructions to reproduce the results.

Doubly Trained Neural Machine Translation System for Adversarial Attack and Data Augmentation Languages Experimented: Data Overview: Source Target Tra

Steven Tan 1 Aug 18, 2022
Cognate Detection Repository

Cognate Detection Repository Details This repository contains the data for two publications: Challenge Dataset of Cognates and False Friend Pairs from

Diptesh Kanojia 1 Apr 26, 2022
Auditing Black-Box Prediction Models for Data Minimization Compliance

Data-Minimization-Auditor An auditing tool for model-instability based data minimization that is introduced in "Auditing Black-Box Prediction Models f

Bashir Rastegarpanah 2 Mar 24, 2022
Research code for Arxiv paper "Camera Motion Agnostic 3D Human Pose Estimation"

GMR(Camera Motion Agnostic 3D Human Pose Estimation) This repo provides the source code of our arXiv paper: Seong Hyun Kim, Sunwon Jeong, Sungbum Park

Seong Hyun Kim 1 Feb 07, 2022
Deep Learning Theory

Deep Learning Theory 整理了一些深度学习的理论相关内容,持续更新。 Overview Recent advances in deep learning theory 总结了目前深度学习理论研究的六个方向的一些结果,概述型,没做深入探讨(2021)。 1.1 complexity

fq 103 Jan 04, 2023
Automatic deep learning for image classification.

AutoDL AutoDL automates machine learning tasks enabling you to easily achieve strong predictive performance in your applications. With just a few line

wenqi 2 Oct 12, 2022