Privacy, Fairness, Accountability and Transparency in Computer Vision

The advances in computer vision research have transformed the way people work and think. Deep learning techniques has outperformed classical machine learning and exceeded human performance, demonstrating the potential to translate computer vision in critical real applications. Nevertheless, applying these techniques broadly in privacy sensitive domains is met with significant hurdles, including ethical considerations, safety, and privacy issues, all of which must be thoroughly considered and resolved prior to widespread adoption. Furthermore, the ethical consideration of employing these technologies to continuous monitoring has been underestimated, since signatures of biometrics can be revealed even when subjects are not directly identifiable. This workshop invites outstanding works on this technically challenging domain to reveal threats and ethical issues and propose solutions.