Content-based annotation and classification framework

Abstract

Unprecedented amounts of digital data are becoming available nowadays, but frequently the data lack some semantic information necessary to effectively organize these resources. For images in particular, textual annotations that represent the semantics are highly desirable. Only a small percentage of images is created with reliable annotations, therefore a lot of effort is being invested into automatic image annotation. In this paper, we address the annotation problem from a general perspective and introduce a new annotation model that is applicable to many text assignment problems. We also provide experimental results from several implemented instances of our model.

CLAN Photo Presenter

“…Furthermore, two postprocessing methods are available to deal with anomalies such as single-member clusters or outlier images. The annotation phase exploits the MUFIN Image Annotation software [1].…”

“…In annotation, the average precision of the MUFIN Image Annotation tool is about 60 % [1]. The group-annotation quality depends on the coherence of a given group of images and the number of representatives used for annotation processing.…”

Visual Concept Ontology for Image Annotations

“…In context of search-based image annotation, the WordNet is very useful for identifying synonymous words and discovering relationships between concepts [1,10]. However, the WordNet relationships only cover linguistic dependencies, which are not satisfactory for image metadata analysis -for instance, there is no relationship linking "roof" and "house" although these words are clearly semantically related.…”

“…As we discussed in the introduction, knowledge bases that provide information about visual concepts and their relationships are much needed for search-based annotation. During the development of a general-purpose annotation system [1], we discovered that none of the existing resources fulfills the specific needs of such application. Therefore, a new Visual Concept Ontology was created for the specific needs of image content description.…”

“…on the web. Specifically, content-based retrieval is applied to obtain images that are visually similar to I, and their metadata is used for reasoning about the probability of concepts from V [1,10]. The whole process is schematically depicted in Figure 1.…”