The York Urban Line Segment Database consists of 102 images of urban environments, 45 of which are indoor scenes and 57 are outdoor scenes. The images are mainly scenes of York University campus and downtown Toronto in Canada, with a size of 640 x 480 pixels, and were taken by a calibrated Panasonic Lumix DMC-LC80 digital camera.

Each image in the dataset has been manually annotated to identify sets of line segments that satisfy the “Manhattan assumption” (Coughlan & Yuille 2003). The Manhattan assumption refers to sets of line segments that conform to a 3D orthogonal framework for an urban environment.