This study investigates the utility of damage detection products, derived using object-based image analysis, for use in preliminary damage assessment for hurricane disasters, the degree to which these products detect damage to residential structures, and the viability of a point-polygon accuracy assessment as an alternative to a traditional pixel-based accuracy assessment. Four datasets, commonly utilized for change detection studies, were created for the Galveston, TX study area to assess grade III damage and the Bolivar Peninsula study area to assess grade IV and V damage. These input datasets were segmented, classified, and their results aggregated to form damage detection products. The final product derived from the principal components composite dataset was the most accurate at detecting grade IV and V damage with an overall accuracy of 0.86, an overall kappa of 0.78. The damage class had a producer's accuracy, user's accuracy, and per-class kappa values of 0.82, 0.87, and 0.75, respectively using a pixel-based assessment. An accurate grade III damage detection product was not derived. Implementing the point-in-polygon-based accuracy assessment was problematic and proved not to be an object-based alternative to a pixel-based accuracy assessment. The PCA composite final product of grade IV and V damage may not be sufficiently accurate to delineate this type of damage at the household level; however, it may be accurate enough for visualizing severe hurricane damage across a region, estimating medium term assistance needs, and in instances where ground access is not possible, providing data about damage households.