The project has a duration of three years and was started in 2009 within the Interreg IV Italy-Austria program funded by the European Union. It aims at building a smart network of audio and video sensors that can automatically localize and detect different types of events in a specific environment. SRSnet, an acronym meaning “Smart Resource-aware multi-Sensor NETworks”, involves the participation of Italian and Austrian partners featuring a long experience in research areas such as computer vision, signal processing, pervasive computing, artificial intelligence, Web technologies, and self-organizing network systems.
Among these, Lakeside Labs is responsible for the coordination and the financial management of the project. The Alpe-Adria University of Klagenfurt takes part in the project with the Pervasive Computing Group and the Transportation Informatics Group. The Italian partners participating to the project are two spin-off companies of the University of Udine: Eye-Tech, specialized in real time Computer Vision and Signal Processing, and Infofactory, a company experienced in Semantic Technologies and Web Search and Monitoring.
One objective of the project is the design and development of a working prototype, currently tested at the National Park Hohe Tauern, Carinthia, chosen as test environment place for a rich variety of realistic case studies. Through this work, the park staff can monitor several factors such as the flow of visitors who annually comes to the park, the behavior of different animals and various activities in the main points of interest.
A further goal of SRSnet is the construction of a multimedia archive including all the audio and video clips related to detected event, which allows to produce statistics and to discover new relevant knowledge extracted from the analysis of the events detected over a period of time.
The benefits of this innovative kind of surveillance system are especially significant in applications for safety and accident’s prevention, since they allow the automatic identification of specific disturbing factors or malfunctions and the subsequent automatic alert of the staff that can timely take the most appropriate actions.
The development of this prototype (still ongoing) has several technical challenges arising from the need to install the system in an outdoor space and keep it running for long periods of time during the year. Among the main issues to be solved there is the problem of self-management of energy supply, provided by solar panels that power cameras and microphones. Another hard problem approached is concerned with the detection and the geographical and temporal localization of significant events happening in the outdoor environment. Another difficulty arises from the need to cover a wide area like a natural park: as soon as a camera identifies a significant event, it focuses on the action that is taking place, and so the other cameras must reconfigure automatically their orientation in order to continue to cover the entire park area or to follow the detected event if it has moved to another location (for example an person running).
During the sessions and events organized at the AVSS conference, researchers from all partners had the opportunity to illustrate the results produced so far in the project and to discuss about specific technical challenges, application areas, and future goals.