Brazilian project uses AI to identify and cover female nudity

An algorithm trained by a group of Brazilian researchers promises to help platforms develop more sophisticated anti-pornography parent filters than today. Four students at the Pontifical Catholic University (PUC) of Rio Grande do Sul have been using artificial intelligence to recognize photos of nude women and replace them with virtual bikinis.

The project uses a deep-learning, machine-learning neural network powered by some 2, 500 images of nude or bikini-clad women. It is from this base that the machine uses its references to detect pornography and edit the image. The idea is to propose a type of “sieve” that does this work quickly and presents the result before the original content is viewed by the user.

“Ideally, it would be desirable to have a method that censors sensitive content non-intrusively so that the user does not notice the nudity” that was in the image before the filter acted, says an excerpt from the study. possibility of scanning also for male nudity.

As you can see, the system still doesn't cover completely accurately and in some cases the coverage is half out of the body - but this should work better over time. In addition to this survey, the PUC-RS AI nucleus has been training algorithms to identify diseases in medical examination photos, as well as recovering lost material involving image manipulation.

***

Do you know the Mega Curioso newsletter? Weekly, we produce exclusive content for lovers of the biggest curiosities and bizarres of this big world! Register your email and do not miss this way to keep in touch!

Brazilian project uses AI to identify and cover female nudity via TecMundo