A workshop on multimedia will be organized along with CVIP-2017 comprising of lectures from eminent researchers from the field.
September 9, 2017:
09.30 am - 11.00 am: Talk 1 (Speaker will be updated soon)
11.00 am - 11.30 am: High Tea
11.30 am - 1 pm: Talk 2 (Speaker will be updated soon)
1 pm - 2.30 pm: Lunch
2.30 pm - 4 pm: Talk 3 (Speaker will be updated soon)
4 pm - 4.15 pm: Tea Break
4.15 pm - 5.00 pm: Panel Discussion (Topic and Delegates will be updated soon)
Multimedia and Multimedia Systems Research:
Medium is defined as the means of communication such as video, audio and text. With the evolution of technology, people started using multiple mediums together for communication, hence emerged the term multimedia. In Multimedia Systems, we extract information from multiple mediums, such as video, audio and text, to design and develop systems for specific applications. Such systems involve exploiting signals from multiple sensors such as camera, audio sensors, motion sensors, GPS, etc. While video is the main and a very rich source of information, researchers are actively exploiting additional sensors such as audio, motion sensors, GPS, etc. The goal is to look at the multiple modalities collectively and accomplish tasks that are hard or inefficient to accomplish with a single modality. Modern Multimedia Systems also leverage on abundant information available on the web, e.g. online social networks, news sites, etc., and on metadata of media such as location, sensory parameters and user preferences.
Some of the main components in Multimedia research are:
- How to synchronously capture multimedia data?
- How to efficiently compress the multimedia data for streaming and storage?
- How to effectively analyse and fuse multimedia data?
Multimedia data comes in various forms from multiple sources, which are needed to be aligned over the same timeline. This becomes even more challenging when the data is captured at different sampling rates and the features are calculated at varying time windows.
The raw multimedia data, particularly videos and audios, could take up a lot of space. So, we need mechanisms to efficiently compress the multimedia data without compromising the desired quality for streaming and storing efficiently. Apart from perceptual quality requirements guiding the mechanisms, the approaches can also be designed based on analysis requirements, like in surveillance, the requirement is preservation of features pertaining to it.
This is the most challenging part of multimedia systems. Once we have the data from multiple sources, how do we synergistically combine the data? The sensors could be providing auxiliary information or complimentary information. In each case, what statistical tools should we use to improve the accuracy and efficiency of the overall task? The information fusion can take place at multiple levels, such as feature and decision levels. What fusion strategy would work best in a given scenario?
The challenges lie in innovative and creative use of data analysis techniques to combine and manage information from multimodal data. A few links having description and demos research projects in Multimedia and Multimedia Systems are: