Multimedia Digital Library Search Tools

Summary: Multimedia library search and navigation tools for professional, educational (both for educators and researchers) and home users.

Reference Code: 0205

With the explosive growth of digital content of all kinds, the need for tools to access and understand multi-modal content is more urgent than ever. Despite a decade of intensive research, tools and methods are still in their infancy, and this holds especially for audiovisual material (still images, video, and audio recordings).

The products of this research are three multimedia digital library search and navigation systems targeted at three different user groups: professional users, educators and researchers, and home users. These systems integrate state-of-the-art audio, visual, and information retrieval tools for indexing and retrieving large volumes of multimedia content. The research provides different types of users with new and engaging ways to interact with audiovisual libraries, helping them discover, browse, navigate, search and enrich archives.

Distinctive aspects of the research

This research is opening up traditionally introspective archival practices to innovative tools and practices enabling improved content exploitation.

Potential applications

• Indexing and search systems for large digital multimedia libraries.

• Indexing and search tools for small and medium archives.

• Feature extraction tools and search services for a range of other applications.

Partnership sought

The researchers are seeking a commercial partner for product development. They are also looking at commercial licensing of individual software components.

LinkedInTwitterFacebookShare

Activity Recognition From Wearable Sensors

Summary: Activity recognition technology for sports and other activities, using machine learning in conjunction with smartphones.

Reference Code: 0206

Accelerometers, gyroscopes and other sensors placed on humans provide raw data which, when combined with machine learning, can enable classification of activities. Most current work in this field is limited to domestic tasks such as walking up stairs, brushing teeth and sitting down. This new research is extending the technology into more challenging domains, in particular, sport.

Previously, machine learning methods have been successfully deployed in computer vision, Internet search, stock market analysis and other areas. Now they have reached a level of sophistication such that they can be used in recognising and classifying many types of human activity.

Distinctive aspects of the research

Currently, most research in complex activity recognition uses multiple expensive sensors, prohibiting many people from using the technology. By contrast, this research uses smartphones which have sensors embedded in them allowing for activity identification. The great advantage is that smartphones already have a wide user base and ownership is expected to grow rapidly still.

In comparative studies the researchers found their smartphone classification approach to be comparable to systems using expensive sensor sets.

Potential applications

• A mobile phone app that allows a team member to see his/ her own stats at the end of a game.

• Coaching tools, providing valuable information to sports coaches on individual performance.

• Health applications, allowing patients to monitor motor improvements following therapy/ rehabilitation.

Partnership sought

The researchers are seeking a commercial partner for some refinement of techniques and commercial product development.

Resources and information

Video: http://www.redbrick.dcu.ie/~deadlock/DIR/Work/Hockey/Demo-Edmond-Hockey.avi

LinkedInTwitterFacebookShare

3D Interactive Museum Exhibition

Summary: Virtual museum/ gallery using off-the-shelf tools to enable interaction with cultural/ historical artifacts in a realistic 3D museum environment.

Reference Code: 0203

Many of the world's leading museums and galleries are devoting vast resources to digitising their collections, producing flawlessly detailed 3D models of priceless artifacts and informative multimedia histories to accompany them.

Exploiting this new treasure house of historical/ cultural information, the 3D Virtual Museum enables immersive interaction in an engagingly photo-realistic 3D environment. In this meticulously recreated museum space, neither distance nor disability can prevent the connected citizen from engaging with the world's cultural treasures and artistic masterpieces.

Distinctive aspects of the research

A key feature of the 3D Virtual Museum is that it was developed largely using low-cost or freely available off-the-shelf tools, putting the recreation of digital exhibitions within reach even of modest budgets.

In a recent collaboration with a major national museum of archaeology, the researchers integrated the virtual environment with Microsoft Kinect to allow for hands-free navigation/ interaction.

Potential applications

• Digital interactive installations at museums/ galleries enabling visitors to explore reconstructions of exhibitions that are closed or no longer running, or to see and handle objects that are archived or too fragile for physical display.

• Virtual school field trips to enable pupils, whatever their location, to visit world-famous museums without the costs of travelling.

• Systems to enable disabled people to enjoy museum experiences that would be difficult or impossible in reality.

Partnership sought

The creators of 3D Virtual Museum are seeking a commercial partner for further research and development, especially in the automation of capture processes to facilitate content creation by museum staff.

LinkedInTwitterFacebookShare

Web3D 2013 Conference in San Sebastian, Spain

The 18th ACM International Web3D Conference will take place in San Sebastian, Spain on 20-22 June 2013. Web3D is a major annual event that brings together researchers, developers, entrepreneurs, artists and content creators to explore new developments in 3D web and multimedia technologies, interactive 3D graphics, information integration and usability in contexts from mobile devices to immersive environments.

Focal areas of Web3D 2013 are: (i) collaborative authoring and delivery of educational, multimedia, industrial, medical, gaming and GIS contents; and (ii) 3D web innovations and applications for e-health, transport, manufacturing and tourism.

Papers are invited on topics including, but not limited to:

  • Modeling, processing, analysis and rendering of complex geometry, structure, and behaviours
  • 3D search, shape matching and indexing
  • Rendering algorithms and standardisation and visualisation of large data sets
  • Interaction methods for online 3D content
  • Interactive 3D graphics for mobile devices
  • Mixed and augmented realities (including standardisation aspects)
  • Agents, animated humanoids, and complex reactive characters
  • Remote rendering and streaming
  • Stereo and multi-view visualisation of 3D graphic interfaces
  • High-performance 3D graphics for distributed environments, tele-presence, tele-operation, and 3D broadcasting
  • Web, multimedia and standards integration and interoperation.

The deadline for submission of papers is 11 March 2013. For further details see the Web3D 2013 conference website.

LinkedInTwitterFacebookShare

H.264/AVC to HEVC Video Transcoder

Summary: Faster, more efficient techniques for converting H.264/AVC video bitstreams to high-quality HEVC bitstreams.

Reference Code: 0406

The new standard for video coding, called High Efficiency Video Coding (HEVC), has just been ratified by the ITU-T and ISO/IEC as successor to the highly successful H.264/AVC. Therefore the motivation for a H.264/AVC to HEVC transcoder is twofold:

(i) to be ready to promote inter-operability for legacy video encoded in H.264/AVC format, when new applications using HEVC emerge;

(ii) to be able to take advantage of the superior rate-distortion performance of HEVC in migrating the abundant existing video in H.264/AVC format.

Distinctive aspects of the research

Transcoding to the HEVC standard presents new challenges not met by existing video transcoding techniques. The researchers have developed techniques for efficiently converting H.264/AVC bitstreams to HEVC bitstreams, enabling much faster transcoding while maintaining good quality in terms of rate-distortion. The techniques can reduce the bitrate by up to 50% while retaining the quality of the original video.

Potential applications

Tools for efficiently converting large video databases to the new HEVC standard.

Partnership sought

The researches seek a commercial partner for further research and product development. Currently, the techniques are able to exploit some information in the H.264/AVC bitstream (e.g. motion vectors and transform coefficients), but use of other information could further speed up transcoding. The research effort for that is estimated at 1-2 years.

LinkedInTwitterFacebookShare