There were several workshops on the last day of the program at ACM Multimedia. I attended parts of three. I gave the keynote talk at the workshop on Video Surveillance. My talk was on Observation Systems. This talk is available here.
I find that the technology has now evolved to the stage that we can think of implementing large scale systems to understand what is happening or happened in a microcosm of interest. And with time this microcosm could be scaled to the whole world, as may be required. Obviously, this requires integrating many technologies that are currently either not developed or developed but are not ready to work with other components. The biggest hurdle in the implementation of these systems is the current object oriented design methodology. We will require thinking in terms of events and consider objects as participants in events. Thus, in this thinking events and objects will co-exist. I am trying to build a ‘global lab’ by pulling together a team of relevant active researchers from different countries, as may be required. This is becoming very exciting. I already received very good support from several prominent groups and researchers.
I also presented research being done by PilHo Kim at Georgia Tech in a workshop (Continuous Archival and Retrieval of Personal Experiences) related to eChronicles organized by Jim Gemmel of Microsoft. PilHo has been developing this interesting approach in which the metadata related to different components and processes used in multimedia systems is being captured so context of data acquisition as well as processing done can be explicitly represented and used. This will help not only in processing in similar situations but also in retrieval of appropriate past data that may have been acquired. This is being done in the context of eChronicles. This presentation is available here.
A panel in the workshop on Multimedia Information Retrieval addressed the issue of the relevance of multimedia in search. Panel agreed that this is now the most exciting area of search. It was interesting to see, however, that they believed that earlier approaches (taxonomy, automatic analysis) do not work and folksonomy maybe the way to go. As a member from audience I asked about the success of folksonomy, but did not get a satisfactory answer. My views on this topic will be addressed in a separate post on this topic.
Now I am back from the conference – this is being typed in my flight and will be posted on getting back home. Overall, Tat Seng Chua and his team organized one of the best organized ACM Multimedia conference. A very enjoyable and productive 4 days in Singapore.