ITEA is the Eureka Cluster on software innovation
ITEA is the Eureka Cluster on software innovation
ITEA 4 page header azure circular

Diffusion Model Based Facial Anonymization in TinyML setup

Project
20219 IML4E
Type
Enhancement
Description

Slightly improved generated image quality and much improved generation performance in terms of wall clock time on TinyML devices. Github: https://github.com/balazsmorv/facediffusion

Contact
Balázs Tibor Morvay
Email
morvayb@edu.bme.hu
Research area(s)
Anonymization, Privacy
Technical features

Machine learning pipeline consisting of a diffusion implicit model and an image super resolution model, to anonymize faces in images.

Integration constraints

Python environment, specified packages

Targeted customer(s)

Dataset curators that are looking for ways to anonymize a dataset

Conditions for reuse

MIT Licence

Confidentiality
Public
Publication date
02-05-2024
Involved partners
Budapest University of Technology and Economics (HUN)