High level visual scene classification using background knowledge of objects

This paper introduces a novel and simple approach of high-level scene classification. Knowing that objects are the essence of any given scene, the proposed method uses them to construct a well-structured background knowledge, which is composed of ranking functions and a statistical collection, in order to support the scene classification process. Since not all objects are relevant, only the most salient ones are identified and used in computing the appropriate scene category. To prove the efficiency of the proposed method, experiments are conducted on state of the art datasets: MIT Indoor, SUN900, SUN2012, SUN397 and LabelMe+.Comparisons with other methods were also introduced. The obtained results are reported and discussed.

Automated summary

This paper introduces a novel and simple approach of high-level scene classification utilizing objects to construct background knowledge. The most salient objects are identified and used in computing the appropriate scene category. Experimental results are reported and discussed to prove the efficiency of the proposed method.