of the SPATION project

Ended January 2004


Consumer electronics devices will soon have a huge storage capacity, enormous processing power, and offer high flexibility. Additionally high speed wired or wireless networks provide the opportunity to interconnect these devices. This results in a home network with tremendous possibilities. Such a system is difficult to design, but the real challenge is to make it easy to use. In Spation we focus on storing and retrieving information in the home. Many of the devices in the home will offer the possibility to store hundreds of hours of video, thousands of songs, and ten thousands of photographs. We need to find solutions to help the user to move, retrieve, and organize this information. From a technical perspective this means we need to:
  • Find solutions which create a distributed storage space by interconnecting Consumer Electronics devices.
  • Develop methods to analyze content to support searching.
  • Investigate how meta-data can be handled in a distributed home storage environment in a way that is transparent to the user.
Key features:
  • Finding video content, browsing through keyframes, watching trailers and summaries
  • Searching photographs, annotating, surfing, creating slide shows, transferring to friends
  • Organizing the distributed storage space, moving content from one device to another.
  • Making a play-list to take with you on a portable MP3 player
  • Controlling devices in the home, providing new ways of interaction

Snapshot of photo-browsing on a handheld

Automatic generation of Summaries and Trailers

To quickly get an overview of stored content or to decide whether a recorded TV program is nice to watch, trailers are ideal. Even when a trailer is not provided by the broadcaster the home system can generate one automatically. Using content analysis algorithms the system gathers information about the scenes in the program. The increase of available processing power gives us the opportunity to develop complex algorithms that provide us with more and more information about what is actually happening in the video. These algorithms range from simple methods like explosion detection, to complex face recognition systems to identify actors. Using information from various content analysis algorithms a rule based system selects scenes from the original content and concatenates it to an appealing trailer.

To remember whether you’ve seen a program before or to check whether this was the episode you are looking for summaries are ideal. In this case a pictorial overview consisting of a fixed number of static frames is provided rather than video fragments. The images are selected based on a clustering algorithm. Similar pictures are placed in the same cluster by this algorithm, for example a scene in the living room and a scene outdoor. Then from each cluster a picture is chosen. This results in a selection of pictures that represent the content as well as possible.

Usage Scenario:

Your Photographs, music and videos are stored in several devices in the house, yet to find what you are looking for, getting up from the couch is not necessary.

Using wireless communication the handheld queries the home system to search for your content. Powerful content analysis algorithms make it possible to search for specific content without having to scroll through hundreds of results.

When content is found it can be stored in the handheld to take it with you or it can be directed to any display in the house to enjoy it together.

This project is a 28-months (2001-2004) RTD-project sponsored by the European Commission under the IST-programme.

Contents of the SPATION pages:


How to contact the SPATION project manager

For all other matters: Euro Partners




URL of this website is: http://www.extra.research.philips.com/euprojects/spation/

Last updat by webmaster:  02-Feb-2005.   (c)  SPATION  project 2001-2005