 This paper compares and evaluates two approaches for automatic building detection and localization using high spatial resolution imagery and LiDAR data, thresholding-based and object-based classification. The thresholding-based approach uses LiDAR data to establish minimum height and vegetation presence thresholds, while the object-based approach follows a standard scheme of image segmentation, feature extraction and selection, using decision trees for classification. Quality assessment is performed at two levels, area and object, with building delineation performance evaluated at the area level and accuracy in spatial location assessed at the object level. The results show high efficiency of both methods, particularly the thresholding-based approach when properly adjusted to the urban landscape type. This article was authored by Javier Estornal, Jorge A. Ricio, Tzamen Hermosaula and others.