brighter AI Makes Collection of Visual Data Privacy Research Accessible on Github

22. October 2020

In the current technological boom of the 21st Century, it is hard to go even a day without hearing casual references to things like machine learning, artificial intelligence, and computer vision. Whether it be via security cameras on a commuter train, or an internet advertisement about a smart home hub, most people these days are well familiar with the aforementioned concepts. What is too commonly overlooked is the subsection of privacy concerns that arise as a result of these technological innovations. From speech analysis to facial recognition, AI-driven technologies are inherently built to process private and personal data. Going about an everyday routine in the city, it is safe to assume that personal information — audio and video information — has been processed within some form of machine learning application at least once. The question then arises, who is responsible for the collection of data being processed; the application user, the company who owns it, or the developers who created it? In reference to the ever-expanding AI-driven data analytics revolution, the Research and Development teams at brighter AI Technologies GmbH, have proposed that “it is a social responsibility to protect individuals’ privacy linked to [their] data,” and have made the Awesome Privacy Papers repository publicly available on Github, in an attempt to bring more awareness and education to the topic at hand. 

The repository is a collection of scientific and academic studies and papers concerning AI-based data analytics and the privacy discussions and solutions that follow. Ignacio Fiedler, Senior Research Scientist at brighter AI, says that the initial idea for the repository branched from the growing collection of resources gathered by the research team in the process of developing the company’s own AI-based video anonymization solutions. The repository has been a team-driven project from the start, and it seems that the creators aim to broaden the spectrum of contributors. Fiedler says that it “would be great to have permanent collaborators that help us to expand and share about this specific topic.” On the repository’s ‘README’ page, the team states, “We care deeply about privacy. To strengthen our knowledge in this field and understand how it relates to visual data, we do constant research in the latest scientific works about this topic.” The mission behind the Awesome Privacy Papers is very clear; to help the machine learning industry evolve into a more privacy-conscious field. 

At its core, the Awesome Privacy Papers aims to increase awareness and provide educational resources for everyone in the machine learning industry. From one perspective, the repository serves simply as a body of resources, collected by professionals in the industry, to demonstrate the importance of the topic. Additionally, the Awesome Privacy Papers are here to provide a starting point for research surrounding privacy protection solutions, especially in areas of the field where traditional license plate and facial redaction techniques, like blurring, significantly reduce the value of the collected data. With more awareness around the legal obligations of organizations to protect user data, and the discussions arising about the additional moral obligations, there is an increasing demand for implementations that allow anonymized and privacy-oriented data collection, while still fulfilling the original objective. Technologies like brighter AI’s Deep Natural Anonymization are the way of the future for organizations that need to utilize personal data without exposing identifiable and private information. The Awesome Privacy Papers are a result of the development of brighter AI’s anonymization technologies and are being published with the intention of facilitating further innovations in the industry. With the repository being publicly available, other developers and persons in the industry are also encouraged to contribute to the collection of resources and to help increase recognition around the importance of the AI and data privacy discussion.

If you are interested in testing brighter AI’s image and video redaction solution for faces and license plates, feel free to use our new demo.

Frederic Scheer
IT Projects
frederic@brighter.ai