Google Shares Open-Source Tool That Automatically Blurs Videos For Privacy
By Alexa Heah, 23 Dec 2022
It’s not uncommon for a person to find themself on a side of the internet they wanted no part in.
Google has released an open-source machine-learning tool, dubbed ‘Magritte’, that will help automatically blur portions of images or videos when they appear on the screen.
This blur effect can be applied to a variety of objects, including human faces, license plates, or even tattoos. Plus, it allows developers and researchers to be able to analyze data without revealing an individual’s personal information.
According to Google, the algorithm is able to anonymize subjects and items with “high accuracy,” and it could be a valuable resource in digital privacy. Typically, object blurring isn’t efficient as it requires a computational intensity that disrupts the user experience.
However, Magritte uses low computational resources to detect objects within images and videos, automatically shielding them as soon as they appear.
The tech giant posits that the tool could be especially helpful to video journalists worldwide who need to anonymize interviewees, or videographers who waste time blurring objects out in videos.
Additionally, the company is rolling out another privacy-enhancing technology (PET), called the Fully Homomorphic Encryption Transpiler, that is a code to allow developers to encrypt data in a set, so that programmers can work on it without being able to access any personal information.
“As we’ve said before, we believe that every internet user in the world deserves world-class privacy… We’re excited for new testing and feedback on our open-source PETs and look forward to releasing more updates in 2023 and beyond,” concludes Google.