When multimedia meets surveillance and forensics in people security

Abstract

When new disciplines emerge, and their commercial applications arise, the market invents new terms and definitions. This is the case of new fields related to surveillance and forensics for people security, which are adopting multimedia, computer vision, content-based retrieval technologies massively. Thus new tools of "intelligent video analytics", "VCA" ("video-content-analysis"), "smart surveillance", "smart forensics" etc. are invading the Web and the ICT market and many research projects in this area are spreading worldwide. This talk aims at presenting the research advances in multimedia and related technologies for people security, that are spreading in applications of real-time video surveillance and off-line analysis of video footage for forensics purposes. In both contexts, two aspects must be taken into account: the data management and the data analysis. In the former aspect, since privacy, legal and security issues are involved, all technology advancements in reliable network transmission, watermarking, secure storing and delivering, transcoding and so on are explored and tailored to the application. In the latter aspect, the research in computer vision, audio analysis, sensor fusion, content-based retrieval, multimedia data mining and metadata analysis are merged with the common goal to extract in real-time or in a very fast way (to cope with the huge amount of data) all the possible knowledge about the scene evidence, the people aspect and the people behavior. This talk, after a brief overview of the state-of-the-art, will focus in particular on this second aspect and in particular on the advances and the challenges of research in people detection, people action and activity analysis, behavior and event detection also in crowd scene covered by forest of cameras. Some recent projects of Modena's ImageLab will be presented together with some results in sensor fusion for people identification and tracking.