Urban 3D maps are desirable to be visually realistic, geographically accurate, and frequently updatable. However, 3D map generation utilizing specialized hardware, such as a LiDAR sensor and an aerial camera, is costly and intended for professional creation of 3D maps. In this paper, we propose a dashcam-based method for creating a 3D map, where 3D building models are automatically generated from dashcam videos and augmented to an open geospatial platform, specifically OpenStreetMap. Since a dashcam captures only a partial side of buildings, we utilize an image-to-3D generation method to reconstruct whole 3D building models and propose an automatic pipeline for globally aligning the 3D building models, the dashcam poses, and the geospatial platform. Because dashcams are inexpensive and available to end users, our method offers a low-cost and widely applicable solution to 3D map generation. Experimental results demonstrate that our method succeeds in generating and augmenting 2,580 buildings for four areas of the map.