AIC Adaptive Information Cluster AIC Home PageContact UsSite Map
   
     
Print a Printer Friendly Version of this PageBack One PageBookmark this Page
Go to Home PageContact UsView Site Map
Log In
   
   
 

Pedestrian Traffic Sensing

Lead: Professor Alan Smeaton;
Dr. Noel O’Connor
Collaborators: MERL and Dublin Corporation

This programme deals with automatic detection and counting of pedestrians using 3D stereo images. This research is motivated by the strong need to automate the management of both traffic and pedestrian flow by intelligent traffic control systems. Its aims to develop techniques to provide dynamic pedestrian flow data, such as the number and nature of pedestrians waiting to cross the road at a given time instant.

 

 
 

Vehicular Traffic Sensing

Lead: Professor Alan Smeaton;
Dr. Noel O’Connor
Collaborators: MERL and Dublin Corporation

Advanced traffic management systems rely on traffic sensors to provide vehicle detection and classification, incident detection and automatic traffic surveillance. While existing inductive loop detectors give information concerning vehicle presence, other parameters such as traffic density and speed must be interpreted from the measured data and this may not have sufficient accuracy for some applications. Furthermore induction loops can be degraded over time and are relatively expensive to install. This project aims to investigate a reliable and cost-effective acoustic vehicle detection and tracking system.

Top of Page

 
 

Using Past Web Queries to Adaptively Inform Future Web Queries

Lead: Professor Alan Smeaton;
Professor Barry Smyth;
 

The team at DCU is building a locally distributed search engine to support searching on a locally-stored collection of 95,000,000 web pages. This will allow a large throughput of local searching for various projects that need to local, fast, large-scale corpus information. In parallel, the team at UCD has developed a technique for web searching which sits on top of a conventional search utility and uses logging of user searches and results to inform and re-rank web pages in subsequent web searches. To date this work has concentrated in domain-specific narrow search areas. The two research teams are collaborating to build a combined local, distributed search utility on top of which is an analysis of past queries to adaptively inform future web queries. The work will be benchmarked against other techniques for searching large web collections in the annual TREC activity, coordinated by the National Institute of Standards and Technology in the US.

Top of Page

 
 

Wireless Sensor Networks

Lead: Professor Dermot Diamond;
Professor Gregory O’Hare
Dr. Rod Shepherd

Collaborators: MERL

This project aims to develop wireless sensor networks that incorporate chemical sensors for feedback on environmental conditions. Initial work will focus on the implementation of such systems using commercial wireless platforms, with the long term goal being to develop and implement our own hardware. The proposed systems are to be self-organising in nature, resulting in truly adaptive, autonomous sensor networks. Our vision is to deploy these in a range of real field trials, where low-power environmental monitoring is required.

Top of Page

 
 

Multi-modal data capture for Audio-Visual Content description

Lead: Noel O'Connor

 

The goal of 'object extraction' has long been the holy grail of audio-visual processing and yet, extensive research using standard video has not led to significant progress. This project hopes to address this challenging research task by using multiple input sources, such as stereo cameras and infrared imagers, to gather more information about a scene and thus make the problem tractable.

Multi-modal data capture for Audio-Visual Content description

Top of Page

 
   
 
Home : About Us : Research Themes : Projects : Events & Publications : Resources : Contact Us : Site Map
Copyright Adaptive Information Cluster (AIC) 2004 & Its Licensors : Designed & Built by Midi Multimedia