CAVIAR - Context Aware Vision using Image-based Active Recognition The project adopts an appearance-based approach to visual recognition. The project addresses the scientific question: Can rich local image descriptions from foveal-like image sensors, selected by a hierarchal visual attention process and processed using task, scene, function and object contextual knowledge improve image-based recognition processes? This clearly addresses questions central to the cognitive vision approach. To investigate this and more specific questions we propose to research methods for l) foveated feature extraction and grouping, 2) integrating feature, object and top-down priming for spatial and temporal attention, 3) representing and recognising objects, contexts and situations, 4) learning representation models from visual evidence and 5) reactive and top-down control of the recognition process. We will integrate the results in a complete closed-loop object and situation recognition system.
Ground Truth Labeled Video Sequences
The CAVIAR project
has collected and hand-labelled ground truth for
81 video sequences comprising about 90K frames.
These sequences include indoor plaza and shopping center
observations of individuals and small groups of people walking, browsing,
window shopping, fighting, meeting, leaving packages behind, collapsing,
entering and exiting shops, etc.
The sequences are labelled (in XML) with both the tracked persons and also
a semantic description of their activities. The data is available as
both mpg sequences and jpg frames. The shopping center data has 2 views for
each frame time, so some wide-baseline stereo tracking experiments
are possible. (A subset of this data was used in the PETS04 workshop.)
The data is at:
http://homepages.inf.ed.ac.uk/rbf/CAVIARDATA1/
and can be used freely by the computer vision community.
(with acknowldgements to EC Funded CAVIAR project/IST 2001 37540).
| ||||||
| ||||||
| Site generated on Friday, 06 January 2006
|
