For effective disaster relief decision-making, responders require extensive and rapid information on the damage situation in affected areas. Areas with unknown conditions pose a high risk of injury, and working on the ground limits the coverage and speed of information acquisition. An alternative is to exploit aerial observations and, in particular, unmanned aerial vehicles (UAVs). UAVs can be rapidly deployed to access remote areas without risking survey teams. Moreover, large-scale disasters impact wide areas, and multiple UAVs are needed to increase coverage without compromising resolution or speed. Of particular importance for evaluation are assets such as hospitals, shelters and essential infrastructures. UAVs can survey such structures to construct three-dimensional (3D) models for inspection.A structure-from-motion (SfM) survey generates 3D models from multiple images. However, most path planning algorithms for SfM focus on points of interest taken from an individual UAV and consider a single structure. Here, we propose a path design method for multi-UAV SfM surveys. By designing flight paths with sufficient overlap and sidelap ratios for all faces of the surveyed objects, more precise 3D models can be constructed than with conventional methods. The fuzzy C-means method is adopted to reduce the UAV flight loads to a uniform minimum to ensure full battery utilization.