About
What is our goal?
In this era of social media, a lot of people post pictures online, consciously sharing their life with friends and followers.
These images however can be used by content classification and retrieval computer systems to extract information the user did not intent to share.
Think of information on location and information about people's appearance and surroundings.
What do we do?
The Pixel Privacy project develops image enhancement technology that helps protecting online user privacy. Specifically, it is dedicated to creating new approaches that invisibly change or visibly enhance images in such ways that it is no longer possible to automatically extract privacy-sensitive information. The methods proposed, PIRE, PerC, and Logit, are three algorithms that change the structure of the pixels, creating this way a new image. This new image won't differ semantically for the human eye, but cannot be used for purposes users did not intend on when posting the picture online.
What is this website for?
On this website you can find examples of the beautiful images we created using image enhancement techniques that also assure user privacy. You can wander around the gallery and be amazed by the beautiful pictures. If you got curious to how we did it, you can have a look at the algorithms page. We described the mechanisms behind each filter using understandable language. And if you really want to dive deep and explore the technical details, you can find links to the published papers on each algorithm and the github repository where you can study the code. In the contact section you can find who to contact if you have questions, want a collaboration or want to have a nice chat on this topic!
