face.report
A new surge in deepfake abuse has turned the internet into a more hostile space for women. With face.report, we hope to add more tools to combat this hostility. Shockingly, more than 90% of deepfake victims are women, who are subject to online sexual harassment or abuse through nonconsensual deepfake pornography.
Mission
face.report searches the web on your behalf and determines if your face has been uploaded without your knowledge. This is an important emerging issue to solve, as AI-generated images and videos can be used with malicious intent to cause major reputational harm. Women especially have been targets of such attacks, and are often belittled, humiliated, and intimidated by the technology.
Our Project
Our service empowers users to expose and take action against those who seek to upload such content containing their likeness by identifying the source of the image using web scraping and facial recognition techniques. With greater computing power we could provide a faster, more comprehensive analysis of the web, and improve the efficacy of the service.
What it actually does
The website begins at a landing page, which educates you on deepfakes and how they can negatively affect people, especially women. To start a search for potential deepfakes, one must provide an image of themselves along with their name. The application then queries google images for their name and downloads each individual image. The image API then checks each image to determine if there are similarities between the uploaded image and the one on the web. It informs users of potential misuse of themselves and allows the user to take appropriate action (such as a takedown or report).
The Process
This project was a major undertaking, with all of us learning technologies we were unfamiliar with. We were able to utilize many useful libraries such as Beautiful Soup, CV2, numpy and pillow. The design of the front end looks particularly polished.
What's Next
With more time we would have liked to refine the process for finding images on the internet, as well as improve the process for transferring the detected matches back to the front end. Another future feature that was planned was a way to flag certain images from the website.
made with love by; Preston Pitzer Alec Neps Brian Lee Gryphyn Ogura
Log in or sign up for Devpost to join the conversation.