When we view a video, regardless of its objective, we can find certain aspects in it that, when we stop, can offer us other complementary information and even reveal other information of a confidential nature that, perhaps, its author or the company did not want to appear. However, it’s too late… or not. Thanks to the Magritte project we can safeguard this information anonymously.
Thanks to Magritte technology and Google, which has shared its open code, it is possible to carry out a blur objects in a video for data protection. This is one of the latest developments that has been announced along with the performance improvements of Fully Homomorphic Encryption (FHE), dedicated to carrying out calculations on encrypted data without accessing the information as collected by EP.
Magritte is included within the so-called PET, technologies with which Google ensures that user information is protected and ensures their anonymity. An aspect for which he has greatly criticized the technological giant in recent times.
And it is that, despite his high cost and its difficult implementation in industries as it requires large computational resourceshopes that companies can take an interest in and exploit this pioneering technology of which there is still much to be exploited.
Google affirms that its objective is to create “a safer ecosystem on the Internet” and hence its attempt to democratize the access of PET by what will become a project of open source where any user will have access to it and you will be able to improve the privacy functions both private and of your company.
Magritte application
But in the practical sense, how can Magritte help? This Google project can help blur objects such as videos, tattoos that may appear, posters, car license plates… using machine learning or automatic learning. This one is in charge of detecting objects and applying an automatic blur as fast as it appears in a video. An algorithm that helps professionals not to promote, view and display information or objects of no interest or with controversial content.
At the same time, Google has also announced new Fully Homomorphic Encryption (FHE) transpiler performance improvements, released in 2021. This is a special compiler that converts the source code of one programming language into another equivalent. A technology that allows developers to perform calculations with encrypted data without accessing personally identifiable information.
With this, Google has included new circuit optimizations for the transpiler, reducing its cost and computational time. So the size of the circuit has decreased by 50%, improving the overall speed.
It is a useful technology for financial services industries, health care and governments by offering proven reliability on their data processing, secure and confidential.
Google has also pointed out how the organizations and governments are exploring the use of PETs to address societal challenges, such as the US and UK, which recently hosted a competition to develop PET solutions for financial crime and public health emergencies.
