AI-powered floor plan analysis for feature extraction in automated valuation models

What is it about?

This article explores how analyzing apartment floor plans using image segmentation can make property value estimates more accurate. The study looked at 5,504 apartments sold in Oslo, focusing on balcony size. Using a machine learning model called FloorPlanNet, the researchers measured balcony sizes from floor plan images and compared these with values obtained through text recognition. They found that including balcony size in valuation models improved accuracy, though the effect on price was less significant as the balcony size increased. This research suggests that image segmentation could be useful for evaluating other features in floor plans, potentially leading to more objective property valuations. This approach could provide architects and developers with better insights into how design elements impact property value. Notably, this is the first study to extract features from floor plans for use in property valuation models.

Featured Image