Last week I've been at MobiMedia and EUMOB 2009. The program started with a tutorial from Fernando Pereira about "Mobile Video Coding" starting from good old object-based coding introduced in MPEG-4. The concept of objects in video is there (i.e., production, etc.) but not yet in our house. He still believes in the concept of object-based video coding! In fact, flexible macroblock ordering (FMO) introduced in Advanced Video Coding (AVC) can be used to simulate object-based coding by grouping slices and prioritize them separately.
One keynote was given by Bernd Girod about Mobile Image Matching – Towards Mobile Augmented Reality. The concept is as follows: (1) Take photo on mobile device, (2) extract certain features, (3) transmit feature to server, (4) do processing on server (e.g., search for associated information about what is on the picture), and (5) send reply to device. This safes bandwidth as the picture does not have to be uploaded to the server, only the feature descriptor which is considerably smaller than the picture itself. And, it works - great ... overall interesting work and lot of people were involved in this project (>10).
Another keynote given by Chang Wen Chen was about "Mobile and Networked Video: A Challenging New Era". He introduced True Ubiquitous Video which is Mobile to Mobile that requires low complexity both at the encoder and decoder. Current coding schemes such as (Advanced Video Coding) AVC and Distributed Video Coding (DVC) require either high complexity at the encoder or decoder which fits well with the targeted application area. He proposes some kind of transcoding between DVC and AVC streams within the network. However, this is at a very early stage and it is not sure whether it leads to any success. Furthermore, he calls for semantic adaptation with more intelligence but I'm not sure what this exactly means ;-)
Finally, EUMOB had some nice presentations from various EC-funded projects such as FUTON, ADAMANTIUM, InterMedia, and OPTIMIX.