Automatic Real-Estate Image Analysis for Retrieval and Classification

Abstract Nowadays, the real estate sector is also benefiting from the advance of the technology. Digital images accompany announcements advertising for renting or selling a property. Computer vision and image analysis contribute to the understanding of these images towards better consumer experience. In this paper we show how retrieval and description can be used efficiently to isolate cvasi-duplicate images and identify similar examples to the ones that have drawn viewer’s interest. Three problems related to real-estate imaging are identified and solutions are proposed and evaluated on large databases.