DeepPrivacy within command-line processor: Command-line implementation of "DeepPrivacy: A Generative Adversarial Network for Face Anonymization" using Open Broadcaster Software (OBS) as a virtual camera."
2022 (English)Artistic output (Unrefereed)
Resource type
Software, multimediaAlternative title
Deep Privacy within command-line processor (English)
Physical description [en]
A folder containing 133 folders and 648 files.80.4 MB or 84,395,353 bytes.
Description [en]
"DeepPrivacy" is a fully automatic anonymization technique for images.
DeepPrivacy was originally released by Håkon Hukkelås, Rudolf Mester, and Frank Lindseth as a research paper "DeepPrivacy: A Generative Adversarial Network for Face Anonymization" in 2019 (https://arxiv.org/abs/1909.04538).
We have used https://github.com/hukkelas/DeepPrivacy repository and implemented the code using python and OBS as a virtual camera.
Open the command line on Windows and install the packages using pip-install (see Guide.md).
The result is that anyone can anonymize the faces in any images or videos as deep-fakes or with astropy rectangles in bulk.
Abstract [en]
Command-line implementation of "DeepPrivacy: A Generative Adversarial Network for Face Anonymization" using Open Broadcaster Software (OBS) as a virtual camera."
Place, publisher, year, pages
Stockholm, 2022.
Keywords [en]
Anonymization, Artificial Intelligence, Broadcaster Software, Computer Vision, DeepPrivacy, Face Anonymization, Generative Adversarial Network, GAN, Machine Learning, OBS, Pattern Recognition, Virtual camera
National Category
Computer graphics and computer vision Humanities and the Arts Visual Arts Arts Human Computer Interaction Computer Sciences
Research subject
Artistic practices
Identifiers
URN: urn:nbn:se:uniarts:diva-1275OAI: oai:DiVA.org:uniarts-1275DiVA, id: diva2:1683944
Funder
Stockholm University of the Arts, F422182022-07-202022-07-202025-09-10Bibliographically approved