The safe and efficient operation of highways largely depends on the adequate provision of sight distance. Small unmanned aerial vehicles (UAVs) can be utilized to efficiently complete data acquisition very soon after identifying an issue when searching for potential highway safety risks. A double grid flight is proposed to obtain an adequate three-dimensional (3D) recreation of the road environment, ensuring an unbiased sight distance output. Then, a dense cloud point is derived through a Structure from Motion Multi-View Stereo process. The point cloud is classified to produce both a terrain model, characterized by its resolution, and a 3D-object model, characterized by the maximum edge length of the entities. The resulting road environment model is utilized to calculate sight distance in a geographic information system. The results enabled the detection of accident-prone locations caused by sight distance limitations. Moreover, the impact of the 3D modeling parameters on the results was assessed.