Computer Vision and Machine Learning for Human Rights Video Analysis: Case Studies, Possibilities, Concerns, and Limitations

Abstract

Citizen video and other publicly available footage can provide evidence of human rights violations and war crimes. The ubiquity of visual data, however, may overwhelm those faced with preserving and analyzing it. This article examines how machine learning and computer vision can be used to make sense of large volumes of video in advocacy and accountability contexts. These technologies can enhance the efficiency and effectiveness of human rights advocacy and accountability efforts, but only if human rights organizations can access the technologies themselves and learn how to use them to promote human rights. As such, computer scientists and software developers working with the human rights community must understand the context in which their products are used and act in solidarity with practitioners. By working together, practitioners and scientists can level the playing field between the human rights community and the entities that perpetrate, tolerate, or seek to cover up violations.