The dataset can be used for learning 6DOF camera pose estimation/visual odometry, image matching, and various 3D estimations. The data comes in bundles of matching images the content of the matching pairs show the same physical point while the camera viewpoint shows a large baseline (often >120 degrees). This repository shares a large dataset of street view images (25 million images and 118 million matching image pairs) with their 6DOF camera pose, 3D models of 8 cities, and extended metadata. ![]() Street View Image, Pose, and 3D Cities Dataset
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |