Crowdsourced 3D Mapping: A Combined Multi-View Geometry and Self-Supervised Learning Approach
This addresses the need for efficient large-scale dynamic mapping and autonomous driving by enabling 3D landmark positioning from monocular cameras and GPS, though it is incremental as it builds on existing multi-view geometry and self-supervised learning methods.
The paper tackles the problem of 3D mapping from crowdsourced visual data without assuming known camera intrinsics, achieving an average single-journey relative positioning accuracy of 39cm and absolute accuracy of 1.26m on a traffic sign dataset.
The ability to efficiently utilize crowdsourced visual data carries immense potential for the domains of large scale dynamic mapping and autonomous driving. However, state-of-the-art methods for crowdsourced 3D mapping assume prior knowledge of camera intrinsics. In this work, we propose a framework that estimates the 3D positions of semantically meaningful landmarks such as traffic signs without assuming known camera intrinsics, using only monocular color camera and GPS. We utilize multi-view geometry as well as deep learning based self-calibration, depth, and ego-motion estimation for traffic sign positioning, and show that combining their strengths is important for increasing the map coverage. To facilitate research on this task, we construct and make available a KITTI based 3D traffic sign ground truth positioning dataset. Using our proposed framework, we achieve an average single-journey relative and absolute positioning accuracy of 39cm and 1.26m respectively, on this dataset.