Ontology-Assisted Object Detection: Towards the Automatic Learning with Internet

Abstract

Abstract. Automatic detection approaches depend essentially on the use of classifiers, that in turn are based on the learning of a given training set. The choice of the training data is crucial: even if this aspect is often neglected, the visual information contained in the training samples can make the difference in a detection/classification scenario. A good training set has to be sufficiently informative to capture the nature of the object under analysis, but at the same time has to be generic enough to avoid overfitting and to cope with new instances of the object of interest. In this paper we follow those approaches that pursue automatic learning from Internet data. We try to show how such training set can be made more appropriate by leveraging on semantic technologies, like lexical resources and ontologies, in the task of retrieving images from the Web through the use of a search engine. Experiments on several object classes of the CalTech101 dataset promote our idea, showing an average increment on the detection accuracy of about 8%.

Statistical Pattern Recognition Meets Formal Ontologies

“…The first methodological results obtained by following this new direction on object recognition can be found in [12,13,2] and on video understanding in [4,3]. Together with papers, practical industrial applications tested as prototypes during the Winter Universiade in Trento 1 are going to be finalized, showing that the potentialities of this new perspective [11] are not limited to a mere theoretical dimension [5].…”

“…The strategy we have applied in some of our early works leverages on semantic relations between words contained in WordNet, chosen with ontologically-driven criteria, coupled with the statistic power provided by Google N-gram to select a set of meaningful text strings related to the text class-label (ex. "bird"), that are subsequently fed into the most common image search engines, producing set of images with high training value [12,13,2]. Our aim was thus that of automatically creating a training set for object recognition by extracting images from Internet.…”