Learning Monocular 3D Vehicle Detection Without 3D Bounding Box Labels
Koestler L, Yang N, Wang R, Cremers D (2021)
Publication Type: Conference contribution
Publication year: 2021
Journal
Publisher: Springer Science and Business Media Deutschland GmbH
Book Volume: 12544 LNCS
Pages Range: 116-129
Conference Proceedings Title: Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Event location: Tübingen, DEU
ISBN: 9783030712778
DOI: 10.1007/978-3-030-71278-5_9
Abstract
The training of deep-learning-based 3D object detectors requires large datasets with 3D bounding box labels for supervision that have to be generated by hand-labeling. We propose a network architecture and training procedure for learning monocular 3D object detection without 3D bounding box labels. By representing the objects as triangular meshes and employing differentiable shape rendering, we define loss functions based on depth maps, segmentation masks, and ego- and object-motion, which are generated by pre-trained, off-the-shelf networks. We evaluate the proposed algorithm on the real-world KITTI dataset and achieve promising performance in comparison to state-of-the-art methods requiring 3D bounding box labels for training and superior performance to conventional baseline methods.
Involved external institutions
Germany (DE)How to cite
APA:
Koestler, L., Yang, N., Wang, R., & Cremers, D. (2021). Learning Monocular 3D Vehicle Detection Without 3D Bounding Box Labels. In Zeynep Akata, Andreas Geiger, Torsten Sattler (Eds.), Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (pp. 116-129). Tübingen, DEU: Springer Science and Business Media Deutschland GmbH.
MLA:
Koestler, Lukas, et al. "Learning Monocular 3D Vehicle Detection Without 3D Bounding Box Labels." Proceedings of the 42nd German Conference on Pattern Recognition, DAGM GCPR 2020 held in parallel with 25th International Symposium on Vision, Modeling, and Visualization, VMV 2020 and 10th Eurographics Workshop on Visual Computing for Biology and Medicine, VCBM 2020, Tübingen, DEU Ed. Zeynep Akata, Andreas Geiger, Torsten Sattler, Springer Science and Business Media Deutschland GmbH, 2021. 116-129.
BibTeX: Download