Catálogo de publicaciones - libros

Selecting suitable topics in order to assess system effectiveness is a crucial part of any benchmark, particularly those for retrieval systems. This includes establishing a range of example search requests (or topics) in order to test various aspects of the retrieval systems under evaluation. In order to assist with selecting topics, we present a measure of topic difficulty for cross-language image retrieval. This measure has enabled us to ground the topic generation process within a methodical and reliable framework for ImageCLEF 2005. This document describes such a measure for topic difficulty, providing concrete examples for every aspect of topic complexity and an analysis of topics used in the ImageCLEF 2003, 2004 and 2005 ad-hoc task.

The aim of the Dublin City University’s participation in the CLEF 2005 ImageCLEF St Andrew’s Collection task was to explore an alternative approach to exploiting text annotation and content-based retrieval in a novel combined way for pseudo relevance feedback (PRF). This method combines evidence from retrieved lists generated using text-based and content-based retrieval to determine which documents will be assumed relevant for the PRF process. Unfortunately the experimental results show that while standard text-based PRF improves upon a no feedback text-only baseline, at present our new approach to combining evidence from text-based and content-based retrieval does not give further improvement.

This paper describes the development of an image retrieval system that combines probabilistic and ontological information. The process is divided in two different stages: indexing and retrieval. Three information flows have been created with different kind of information each one: word forms, stems and stemmed bigrams. The final result combines the results obtained in the three streams. Knowledge is added to the system by means of an ontology created automatically from the St. Andrews Corpus. The system has been evaluated at CLEF05 image retrieval task.

In this paper, we present our participation in the ImageCLEF 2005 ad-hoc task. After a pool of preliminary tests in which we evaluated the impact of different-size dictionaries using three distinct approaches, we proved that the biggest differences were obtained by recognizing named entities and launching structured queries over the metadata. Thus, we decided to refine our named entities recognizer and repeat the three approaches with the 2005 topics, achieving the best result among all cross-language European Spanish to English runs.

When short queries and short image annotations are used in text-based cross-language image retrieval, small changes in word usage due to translation errors may decrease the retrieval performance because of an increase in lexical mismatches. In the ImageCLEF2005 ad-hoc task, we investigated the use of learned word association models that represent how pairs of words are related to absorb such mismatches. We compared a precision-oriented simple word-matching retrieval model and a recall-oriented word association retrieval model. We also investigated combinations of these by introducing a new ranking function that generated comparable output values from both models. Experimental results on English and German topics were discouraging, as the use of word association models degraded the performance. On the other hand, word association models helped retrieval for Japanese topics whose translation quality was low.

This paper regards images with captions as a cross-media parallel corpus, and presents a corpus-based relevance feedback approach to combine the results of visual and textual runs. Experimental results show that this approach performs well. Comparing with the mean average precision (MAP) of the initial visual retrieval, the MAP is increased from 8.29% to 34.25% after relevance feedback from cross-media parallel corpus. The MAP of cross-lingual image retrieval is increased from 23.99% to 39.77% if combining the results of textual run and visual run with relevance feedback. Besides, the monolingual experiments also show the consistent effects of this approach. The MAP of monolingual retrieval is improved from 39.52% to 50.53% when merging the results of the text and image queries.

In this paper, we describe our studies of cross-language and cross-media image retrieval at the ImageCLEF 2005. This is the first participation of our CUHK (The Chinese University of Hong Kong) group at ImageCLEF. The task in which we participated is the “bilingual ad hoc retrieval” task. There are three major focuses and contributions in our participation. The first is the empirical evaluation of language models and smoothing strategies for cross-language image retrieval. The second is the evaluation of cross-media image retrieval, i.e., combining text and visual contents for image retrieval. The last is the evaluation of bilingual image retrieval between English and Chinese. We provide an empirical analysis of our experimental results, in which our approach achieves the best mean average precision result in the monolingual query task in the campaign. Finally we summarize our empirical experience and address the future improvement of our work.

In this paper, we describe our first participation in the ImageCLEF campaign. The SINAI research group participated in both the ad hoc task and the medical task. For the first task, we have used several translation schemes as well as experiments with and without Pseudo Relevance Feedback (PRF). A voting-based system has been developed, for the ad hoc task, joining three different systems of participant Universities. For the medical task, we have also submitted runs with and without PRF, and experiments using only textual query and using textual mixing with visual query.

The CEA-LIST/LIC2M develops both multilingual text retrieval systems and content-based image indexing and retrieval systems. These systems are developed independently. The merging of the results of the two systems is one of the important research interests in our lab. We tested several simple merging techniques in the ImageCLEF 2005 campaign. The analysis of our results show that improved performance can be obtained by appropriately merging the two media. However, an a-priori tuning of the merging parameters is difficult because the performance of each system highly depends on the corpus and queries.

In this paper we report our work using visual feature fusion for the tasks of medical image retrieval and annotation in the benchmark of ImageCLEF 2005. In the retrieval task, we use visual features without text information, having no relevance feedback. Both local and global features in terms of both structural and statistical nature are captured. We first identify visually similar images manually and form templates for each query topic. A pre-filtering process is utilized for a coarse retrieval. In the fine retrieval, two similarity measuring channels with different visual features are used in parallel and then combined in the decision level to produce a final score for image ranking. Our approach is evaluated over all 25 query topics with each containing example image(s) and topic textual statements. Over 50,000 images we achieved a mean average precision of 14.6%, as one of the best performed runs. In the annotation task, visual features are fused in an early stage by concatenation with normalization. We use support vector machines (SVM) with RBF kernels for the classification. Our approach is trained over a 9,000 image training set and tested over the given test set with 1000 images and on 57 classes with a correct classification rate of about 80%.