Real estate images contain a lot of valuable information and are broadly used for marketing purposes in property sales. However, until now, photos of the exterior of a house have not been processed systematically to use the information contained in hedonic pricing models. In this paper we examine to what extent the present information in images can be utilised to derive the influence of the building condition on the value of a house. Thereby, we analyse (i) if we can extract the condition of the building directly from the picture and shows the corresponding expected influence and (ii) to which extend we can improve a hedonic pricing model with the additional information.

We classified 949 single-family houses into four classes according to a uniform condition variable. A regression analysis shows that the created variable together with the year of construction are the key influencing variables. This is in accordance with the theory and the valuation using the cost approach where these variables are also the decisive variables. It can be shown that if the condition variable is added to a baseline model consisting of the square meter of the plot and the building, as well as the price of the land per square meter the new model improves (R-squared: from 0.649 to 0.741, n = 949). This is result is comparable to a model incorporating all variables considered in our examination excluding the condition variable (R-squared: 0.741, n = 360) and yields a better prediction than the baseline model including construction year (R-squared: 0.729, n = 686).