Keynotes
"Civic Multimedia, Crowdsourcing, and the Public Good" Daniel Gatica-Perez (Idiap-EPFL)
Abstract: The engagement of young people in local civic concerns has educational, social, and economic implications. There is an entire open agenda of civic issues that multimedia research can support, spanning media collection, content analysis, and media creation and repurposing.
In the specific context of cities in the developing world, how do youth perceive their urban environment? How do they react to problems like security, accessibility, or waste management? What issues are they more sensitive about? What problems go unnoticed? How can mobile and social tech be used to support young people document these issues, enable self and group reflections, and contribute to community action?
In the talk, I will present a mobile crowdsourcing framework where young people help render visible urban issues that matter to them. Integrating photos and video, online experiments, multimedia analysis, and media creation, the goal is to enable a reflection process through which discussions and proposals to address such issues can emerge. I will argue for the need to think beyond single disciplines, and advocate for the opportunities that emerge from working with communities, both to contribute to the public good and to advance multimedia research in cities.
Bio: Prof. Daniel Gatica-Perez directs the Social Computing Group at Idiap-EPFL in Switzerland. He has worked on human-centered computing for over a decade, integrating research in ubiquitous computing, social media, and the social sciences. His work has been supported by the Swiss National Science Foundation, the European Commission, and companies like Nokia and NTT. His current interests include crowdsourcing and the use of mobile and social technologies for social good.
"Web-Scale Video Content Search" Alex Hauptmann (Carnegie Mellon University)
Abstract: Even though the accuracy of content based video search systems (CBVS) has drastically improved, high accuracy systems tend to be too inefficient for interactive search. Therefore, to achieve real-time CBVS over millions of videos, we perform a comprehensive study on the different components in a CBVS system to understand the tradeoffs between accuracy and speed of each component.
Directions investigated include exploring different low-level and semantics based features, testing different compression factors and approximations during video search, and understanding the time vs. accuracy trade-of. Semantic search in video is a novel and challenging problem in information and multimedia retrieval. Existing solutions are mainly limited to text matching, in which the query words are matched against the textual metadata generated by users.
This talk will contrast approaches for content search both with example videos and without, using only text queries. The system relies on substantial video content analysis and allows for both low-level and semantic search over a large collection of videos. We share our observations and lessons in building such a system, which may be instrumental in guiding the design of future systems for search in video.
Extensive experiments on very large archives consisting of more than 2,000 hours of short videos showed that through a combination of effective features, highly compressed representations, and one iteration of reranking, our proposed system can achieve an 10,000-fold speedup while retaining 80% accuracy of a state-of-the-art CBVS system. Over 1 million videos, our system can complete example-based search in one second with a single core.
Bio: Alexander G. Hauptmann is a Principal Systems Scientist in the Carnegie Mellon University Computer Science Department and a faculty member with CMU’s Language Technologies Institute. His research interests have led him to pursue and combine several different areas: man-machine communication, natural language processing, speech understanding and synthesis, machine learning. He worked on speech and machine translation at CMU from 1984-94, when he joined the Informedia project where he developed the News-on-Demand application. Since then he has conducted research on video analysis and retrieval on broadcast news as well as observational video with success documented by outstanding performance in many video analysis challenges. His current research centers on robust analysis of internet-style and surveillance video as large scale data.
