List of publications

Selected publications

On the Application of Convex Transforms to Metric Search

Scalable similarity search in metric spaces relies on using the mathematical properties of the space in order to allow efficient querying. Most important in this context is the triangle inequality property, which can allow the majority of individual similarity comparisons to be avoided for a given query. However many important metric spaces, typically those with high dimensionality, are not amenable to such techniques. In the past convex transforms have been studied as a pragmatic mechanism which can overcome this effect; however the problem with this approach is that the metric properties may be lost, leading to loss of accuracy. Here, we study the underlying properties of such transforms and their effect on metric indexing mechanisms. We show there are some spaces where certain transforms may be applied without loss of accuracy, and further spaces where we can understand the engineering tradeoffs between accuracy and efficiency. We back these observations with experimental analysis. To highlight the value of the approach, we show three large spaces deriving from practical domains whose dimensionality prevents normal indexing techniques, but where the transforms applied give scalable access with a relatively small loss of accuracy.

MotionMatch: Motion Recognition Technology

MotionMatch is a software technology for recognizing persons according to the way they walk. The recognition process is based on analysis of motion capture data which can be acquired by motion capturing devices, including popular Microsoft Kinect and ASUS Xtion. The acquired data are firstly preprocessed by detecting walking cycles and extracting movement features in form of relative velocities of the specific joints for each walking cycle. Then individual walking cycles can be mutually compared to calculate their similarity. A proposed classification method is finally used to recognize the person who has performed a query motion. The current version of the MotionMatch technology is demonstrated via a web application that allows users to select a query motion and verify whether the technology recognizes the query person correctly.

DISA at ImageCLEF 2014 Revised: Search-based Image Annotation with DeCAF Features

This paper constitutes an extension to the report on DISA MU team participation in the ImageCLEF 2014 Scalable Concept Image Annotation Task as published in [3]. Specifically, we introduce a new similarity search component that was implemented into the system, report on the results achieved by utilizing this component, and analyze the influence of different similarity search parameters on the annotation quality.

Searching for Variable-Speed Motions in Long Sequences of Motion Capture Data

Motion capture data digitally represent human movements by sequences of body configurations in time. Subsequence searching in long sequences of such spatio-temporal data is difficult as query-relevant motions can vary in execution speeds and styles and can occur anywhere in a very long data sequence. To deal with these problems, we employ a fast and effective similarity measure that is elastic. The property of elasticity enables matching of two overlapping but slightly misaligned subsequences with a high confidence. Based on the elasticity, the long data sequence is partitioned into overlapping segments that are organized in multiple levels. The number of levels and sizes of overlaps are optimized to generate a modest number of segments while being able to trace an arbitrary query. In a retrieval phase, a query is always represented as a single segment and fast matched against segments within a relevant level without any costly post-processing. Moreover, visiting adjacent levels makes possible subsequence searching of time-warped (i.e., faster or slower executed) queries. To efficiently search on a large scale, segment features can be binarized and segmentation levels independently indexed. We experimentally demonstrate effectiveness and efficiency of the proposed approach for subsequence searching on a real-life dataset.

All publications



Total number of publications: 2


Responsible contact: test@email.cz

You are running an old browser version. We recommend updating your browser to its latest version.

More info