Dec 23, 2022 12:35:00

Google open sourced ``Magritte'' that can automatically detect and blur faces and objects in videos and publish it on Github

Google has released an open source library `` Magritte '' that can detect objects such as faces and license plates from photos and videos and apply blur. It provides a higher level of obfuscation than simple blurring, and is said to help protect the subject's privacy.

Home |

https://google.github.io/magritte/

GitHub - google/magritte: Mediapipe-based library to redact faces from videos and images

https://github.com/google/magritte

Google Opens Up Tool to Make Your Privates Private
https://gizmodo.com/google-privacy-ai-magritte-1849923870

Google is making its internal video-blurring privacy tool open source | Engadget
https://www.engadget.com/google-is-making-its-internal-video-blurring-privacy-tool-open-source-150052840.html

Magritte is an image editing library for hiding the identity of people in photos and videos. It is possible to detect objects using machine learning and apply blur to arbitrary objects such as license plates and tattoos, which is especially useful for journalists who interview people.

The library can provide higher levels of obfuscation than simple blurring, and you can choose between real-time optimized solutions and offline processing optimized solutions to suit the situation. I can do it.

Magritte is provided as a library built on MediaPipe , so it can be used in various languages and platforms where MediaPipe is available. However, Magritte is in the early stages of development at the time of writing the article, and the accuracy is still being evaluated.

In addition to Magritte, Google has also announced a privacy-enhancing technology called 'Fully Homomorphic Encryption Transpiler'. By using this, developers will be able to perform calculations on encrypted data without accessing personally identifiable information. Google said of the technology, ``We can support industries where strong security assurance regarding the handling of sensitive data is most important, such as financial services, healthcare, and government agencies.''