This is the official code implementation of the "ICIP 2022" paper "GPU-accelerated SIFT-aided source identification of stabilized videos"

• Download the python libraries of Camera-fingerprint ;
• if Camera-fingerprint is not already, reorganize the folders such that PRNU/CameraFingerprint ;
• Download the Reference Camera Fingerprints here;
• at least 9G GPU.

conda env create -f environment.yml

Download Vision dataset here.

nohup python -u main_H1.py --videos PATH_TO_VIDEOS --fingerprint PATH_TO_FINGERPRINTS --output PATH_TO_OUTPUT_FOLDER --gpu_dev /gpu:N >| output_H1.log & 

nohup python -u main_H0.py --videos PATH_TO_VIDEOS --fingerprint PATH_TO_FINGERPRINTS --output PATH_TO_OUTPUT_FOLDER --gpu_dev /gpu:N >| output_H0.log & 

Edit and Run bash runner.sh

You need to edit:

• PATH_TO_VIDEOS changing it with the path to your dataset
• PATH_TO_FINGERPRINTS changing it with the path to your reference camera fingerprints
• PATH_TO_OUTPUT_FOLDER changing it with the path to your output folder
• N chaging it with your GPU ID

Check "GPU-accelerated SIFT-aided source identification of stabilized videos"

If you use this material please cite:

@inproceedings{montibeller2022gpu,
title={GPU-accelerated SIFT-aided source identification of stabilized videos},
author={Montibeller, Andrea and Pasquini, Cecilia and Boato, Giulia and Dell’Anna, Stefano and P{'e}rez-Gonz{'a}lez, Fernando},
booktitle={2022 IEEE International Conference on Image Processing (ICIP)},
pages={2616--2620},
year={2022},
organization={IEEE}
}