Official implementation of the article "Unsupervised JPEG Domain Adaptation For Practical Digital Forensics"

Overview

Unsupervised JPEG Domain Adaptation for Practical Digital Image Forensics

@WIFS2021 (Montpellier, France)

Rony Abecidan, Vincent Itier, Jeremie Boulanger, Patrick Bas

Installation

To be able to reproduce our experiments and do your own ones, please follow our Installation Instructions

Architecture used

Domain Adaptation in action

  • Source : Half of images from the Splicing category of DEFACTO
  • Target : Other half of the images from the Splicing category of DEFACTO, compressed to JPEG with a quality factor of 5%

To have a quick idea of the adaptation impact on the training phase, we selected a batch of size 512 from the target and, we represented the evolution of the final embeddings distributions from this batch during the training according to the setups SrcOnly and Update($\sigma=8$) described in the paper. The training relative to the SrcOnly setup is on the left meanwhile the one relative to Update($\sigma=8$) is on the right.

Don't hesitate to click on the gif below to see it better !

  • As you can observe, in the SrcOnly setup, the forgery detector is more and more prone to false alarms, certainly because compressing images to QF5 results in creating artifacts in the high frequencies that can be misinterpreted by the model. However, it has no real difficulty to identify correctly the forged images.

  • In parallel, in the Update setup, the forgery detector is more informed and make less false alarms during the training.

Discrepancies with the first version of our article

Several modifications have been carried out since the writing of this paper in order to :

  • Generate databases as most clean as possible
  • Make our results as most reproducible as possible
  • Reduce effectively computation time and memory space

Considering that remark, you will not exactly retrieve the results we shared in the first version of the paper with the implementation proposed here. Nevertheless, the results we got from this new implementation are comparable with the previous ones and you should obtain similar results as the ones shared in this page.

For more information about the modifications we performed and the reasons behind, click here

Main references

@inproceedings{mandelli2020training,
  title={Training {CNNs} in Presence of {JPEG} Compression: Multimedia Forensics vs Computer Vision},
  author={Mandelli, Sara and Bonettini, Nicol{\`o} and Bestagini, Paolo and Tubaro, Stefano},
  booktitle={2020 IEEE International Workshop on Information Forensics and Security (WIFS)},
  pages={1--6},
  year={2020},
  organization={IEEE}
}

@inproceedings{bayar2016,
  title={A deep learning approach to universal image manipulation detection using a new convolutional layer},
  author={Bayar, Belhassen and Stamm, Matthew C},
  booktitle={Proceedings of the 4th ACM workshop on information hiding and multimedia security (IH\&MMSec)},
  pages={5--10},
  year={2016}
}

@inproceedings{long2015learning,
  title={Learning transferable features with deep adaptation networks},
  author={Long, M. and Cao, Y. and Wang, J. and Jordan, M.},
  booktitle={International Conference on Machine Learning},
  pages={97--105},
  year={2015},
  organization={PMLR}
}


@inproceedings{DEFACTODataset, 
	author = {Ga{\"e}l Mahfoudi and Badr Tajini and Florent Retraint and Fr{\'e}d{\'e}ric Morain-Nicolier and Jean Luc Dugelay and Marc Pic},
	title={{DEFACTO:} Image and Face Manipulation Dataset},
	booktitle={27th European Signal Processing Conference (EUSIPCO 2019)},
	year={2019}
}

Citing our paper

If you wish to refer to our paper, please use the following BibTeX entry

@inproceedings{abecidan:hal-03374780,
  TITLE = {{Unsupervised JPEG Domain Adaptation for Practical Digital Image Forensics}},
  AUTHOR = {Abecidan, Rony and Itier, Vincent and Boulanger, J{\'e}r{\'e}mie and Bas, Patrick},
  URL = {https://hal.archives-ouvertes.fr/hal-03374780},
  BOOKTITLE = {{WIFS 2021 : IEEE International Workshop on Information Forensics and Security}},
  ADDRESS = {Montpellier, France},
  PUBLISHER = {{IEEE}},
  YEAR = {2021},
  MONTH = Dec,
  PDF = {https://hal.archives-ouvertes.fr/hal-03374780/file/2021_wifs.pdf},
  HAL_ID = {hal-03374780}
}
Owner
Rony Abecidan
PhD Candidate @ Centrale Lille
Rony Abecidan
[CVPR 2021] Region-aware Adaptive Instance Normalization for Image Harmonization

RainNet — Official Pytorch Implementation Region-aware Adaptive Instance Normalization for Image Harmonization Jun Ling, Han Xue, Li Song*, Rong Xie,

130 Dec 11, 2022
FCA: Learning a 3D Full-coverage Vehicle Camouflage for Multi-view Physical Adversarial Attack

FCA: Learning a 3D Full-coverage Vehicle Camouflage for Multi-view Physical Adversarial Attack Case study of the FCA. The code can be find in FCA. Cas

IDRL 21 Dec 15, 2022
Detectron2 is FAIR's next-generation platform for object detection and segmentation.

Detectron2 is Facebook AI Research's next generation software system that implements state-of-the-art object detection algorithms. It is a ground-up r

Facebook Research 23.3k Jan 08, 2023
Code of paper "Compositionally Generalizable 3D Structure Prediction"

Compositionally Generalizable 3D Structure Prediction In this work, We bring in the concept of compositional generalizability and factorizes the 3D sh

Songfang Han 30 Dec 17, 2022
[NeurIPS-2021] Mosaicking to Distill: Knowledge Distillation from Out-of-Domain Data

MosaicKD Code for NeurIPS-21 paper "Mosaicking to Distill: Knowledge Distillation from Out-of-Domain Data" 1. Motivation Natural images share common l

ZJU-VIPA 37 Nov 10, 2022
Resco: A simple python package that report the effect of deep residual learning

resco Description resco is a simple python package that report the effect of dee

Pierre-Arthur Claudé 1 Jun 28, 2022
Api's bulid in Flask perfom to manage Todo Task.

Citymall-task Api's bulid in Flask perfom to manage Todo Task. Installation Requrements : Python: 3.10.0 MongoDB create .env file with variables DB_UR

Aisha Tayyaba 1 Dec 17, 2021
Change Detection in SAR Images Based on Multiscale Capsule Network

SAR_CD_MS_CapsNet Code for the paper "Change Detection in SAR Images Based on Multiscale Capsule Network" , IEEE Geoscience and Remote Sensing Letters

Feng Gao 21 Nov 29, 2022
A no-BS, dead-simple training visualizer for tf-keras

A no-BS, dead-simple training visualizer for tf-keras TrainingDashboard Plot inter-epoch and intra-epoch loss and metrics within a jupyter notebook wi

Vibhu Agrawal 3 May 28, 2021
Volumetric parameterization of the placenta to a flattened template

placenta-flattening A MATLAB algorithm for volumetric mesh parameterization. Developed for mapping a placenta segmentation derived from an MRI image t

Mazdak Abulnaga 12 Mar 14, 2022
Much faster than SORT(Simple Online and Realtime Tracking), a little worse than SORT

QSORT QSORT(Quick + Simple Online and Realtime Tracking) is a simple online and realtime tracking algorithm for 2D multiple object tracking in video s

Yonghye Kwon 8 Jul 27, 2022
Experiments for Fake News explainability project

fake-news-explainability Experiments for fake news explainability project This repository only contains the notebooks used to train the models and eva

Lorenzo Flores (Lj) 1 Dec 03, 2022
Learning to trade under the reinforcement learning framework

Trading Using Q-Learning In this project, I will present an adaptive learning model to trade a single stock under the reinforcement learning framework

Uirá Caiado 470 Nov 28, 2022
PyContinual (An Easy and Extendible Framework for Continual Learning)

PyContinual (An Easy and Extendible Framework for Continual Learning) Easy to Use You can sumply change the baseline, backbone and task, and then read

Zixuan Ke 176 Jan 05, 2023
Voice assistant - Voice assistant with python

🌐 Python Voice Assistant 🌵 - User's greeting 🌵 - Writing tasks to todo-list ?

PythonToday 10 Dec 26, 2022
Official implementation of the NeurIPS 2021 paper Online Learning Of Neural Computations From Sparse Temporal Feedback

Online Learning Of Neural Computations From Sparse Temporal Feedback This repository is the official implementation of the NeurIPS 2021 paper Online L

Lukas Braun 3 Dec 15, 2021
Video Autoencoder: self-supervised disentanglement of 3D structure and motion

Video Autoencoder: self-supervised disentanglement of 3D structure and motion This repository contains the code (in PyTorch) for the model introduced

157 Dec 22, 2022
Automatically erase objects in the video, such as logo, text, etc.

Video-Auto-Wipe Read English Introduction:Here   本人不定期的基于生成技术制作一些好玩有趣的算法模型,这次带来的作品是“视频擦除”方向的应用模型,它实现的功能是自动感知到视频中我们不想看见的部分(譬如广告、水印、字幕、图标等等)然后进行擦除。由于图标擦

seeprettyface.com 141 Dec 26, 2022
Lua-parser-lark - An out-of-box Lua parser written in Lark

An out-of-box Lua parser written in Lark Such parser handles a relaxed version o

Taine Zhao 2 Jul 19, 2022