Carnegie Mellon University
Video Forensics in Human Rights Abuse and War Crimes Investigation: Technology, Law, and Ethics

Video Forensics in Human Rights Abuse and War Crimes Investigation: Technology, Law, and Ethics

The Center for Human Rights Science at Carnegie Mellon University will hold a workshop August 20-21, 2014 that focuses on the collection, preservation, processing and analysis of large volumes of user-generated video to discover digital evidence of human rights abuse and war crimes.  We will pay particular attention to the increasing use of mobile phones, the Internet, and social media by victims, their advocates, belligerents in conflict, and ordinary citizens to document and disseminate audiovisual records of their experiences and perspectives.  Although the workshop will not focus exclusively on a single conflict, the massive amount of audiovisual data from the on-going civil war in Syria will provide the basis for technology demonstrations and conversations about how to collect and store data for future use in justice and accountability activities.

One major purpose of this meeting is to set the stage to harness the tremendous technological and scientific capacity of Carnegie Mellon University (consistently ranked as one of the top schools of computer science in the world) to advance the state of data sciences for human rights and humanitarian uses.  Carnegie Mellon researchers will push the bounds of what is possible with current tools and techniques and devise new methods of data analysis specifically suited to the human rights context.  As such, this is an effort to promote fundamental research in human rights science rather than replicating technology development already being carried out by outstanding organizations like Guardian Project and Witness.  The work that emerges from this meeting will aid human rights advocates, lawyers, academic researchers and decision-makers in their efforts to pour through the thousands upon thousands of hours of video now being produced in fragile societies, conflict zones, and situations of human rights abuse.

The goals of this meeting are to:

  1. provideAmplify and focus an emerging conversation on digital evidence and video forensics in the era of social media and big data analytics
  2. Introduce human rights advocates, lawyers, and academic researchers to state of the art in computer vision, machine learning, language technologies, and data analytics, paying equal attention to the strengths and limitations of these techniques both now and in the long term
  3. discussIntroduce research scientists and technologists to the legal aspects of evidence in human rights abuse and war crimes investigations
  4. Foreground ethical concerns about the collection, storage and analysis of human rights and conflict data and address them before the implementation of data systems, rather than after 

Issues to be addressed

At this workshop human rights advocates, lawyers, academic researchers, computer scientists and technologists will work together to understand and analyze:

Meeting Format

  1. Technical Issues
    1. The current status of video forensics in the context of conflict, human rights abuse, and law enforcement
    2. The state of the art in manual, semi-automated, and automated methods for extracting information from audiovisual sources
    3. The technical challenges associated with organizing and analyzing this data in large databases (including language translation issues) and merging it with other datasets
    4. The extent to which user-generated content provide statistically representative samples of events taking place on the ground
  2. Legal Issues
    1. The relevance of social media (both in terms of actual content and the shear volume of information) on future prosecutions of war criminal and human rights abusers (i.e., accountability and justice)
    2. The extent to which information extracted from user generated content would be deemed relevant and admissible in international tribunals and national courts—e.g., how can its validity and reliability be determined, does it violate rules against self-incrimination, what minimal threshold of relevance must it meet to be introduced into court, etc.
    3. What kinds of computational analyses would be most useful for the legal system (both prosecution and defense) in terms of establishing accountability and achieving justice?
    4. What kinds of analyses would be least helpful for the legal system and might potentially cause long-term harm, whether to the judicial process or to society as a whole?
    5. Are such massive analyses necessary for tribunals, courts, and truth commissions or will more limited analysis (perhaps done by hand) just as effective to prove human rights abuse or war crimes?
  3. Ethical Issues
    1. Is justice served by these kinds of massive analyses?
    2. What are the potential harms associated such computational analysis?
    3. Who should have access to the results of such analyses?
    4. Should the identities of people in the videos be protected in any way?
    5. Should the information extracted from these videos be combined with data from other sources (whether publically available or from government, non-governmental or corporate entities) to produce more fine-grained analyses?
    6. Who should oversee such analyses?
    7. Should these analyses be subjected to Institutional Review Board oversight even though all of the videos and data being gathered are ostensibly in the public domain?
    8. Should the technology required to do such massive analyses of video be open source?


This two-day workshop, to be held at Carnegie Mellon University in Pittsburgh, USA, will build upon experience the organizer has gained putting together similar meetings in the past on civilian casualty recording and estimation in times of conflict; and on the identification of the missing after conflict and disaster.  It will include technology demonstrations, panel presentations and roundtable discussions that allow people to relate their own experiences, concerns and ideas about the issues being presented.  Ample time will be left for informal conversations and networking.

Principal Organizer and Institution

Jay D. Aronson is Associate Professor of Science, Technology, and Society at Carnegie Mellon University. His research and teaching focus is on the interactions of science, technology, law, and human rights in a variety of contexts. He is currently engaged in a long-term study of the ethical, political, and social dimensions of post-conflict and post-disaster identification of the missing and disappeared. This work is funded by the U.S. National Institutes of Health (NIH). Jay is also part of a project that seeks to improve the quality of civilian casualty recording and estimation in times of conflict, is funded by the MacArthur Foundation to evaluate the potential impact of social media and "big data" analytics on human rights fact-finding. He is the founder and director of the Center for Human Rights Science at Carnegie Mellon, which brings together scientists and human rights practitioners committed to rigorous assessment of the state of human rights around the world. The Center encourages interdisciplinary collaboration in order to promote the development and application of scientific methods for collecting, analyzing, and communicating human rights information. The Center also provides technical assistance to individuals and organizations devoted to advancing human rights through consultation, educational programs, and original research.

View the participants »

Related materials »