FCVL: Fourier Cross-View Learning for Generalizable 3D Object Detection in Bird’s Eye View

Xue Zhao, Xinbing Wang, Chenghu Zhou, Qinying Gu, Nanyang Ye

Keywords: Single Domain Generalization，3D Object Detection，Bird’s Eye View

Abstract: Improving the generalization of Birds' Eye View (BEV) detection models is essential for safe driving in real world. In this paper, we consider a realistic yet more challenging scenario, which aims to improve the generalization with single source data for training, as collecting multiple source data is time-consuming and labor intensive in autonomous driving. To achieve this, we rethink the task from a frequency perspective and exploit the cross-view consistency between adjacent perspectives. We propose the Fourier Cross-View Learning (FCVL) framework including Fourier Hierarchical Augmentation (FHiAug), an augmentation strategy in frequency domain to boost domain diversity and Fourier Cross-View Semantic Consistency Loss to facilitate the model to learn more domain-invariant features. Furthermore, we provide theoretical guarantees via augmentation graph theory. To the best of our knowledge, this is the first study to explore generalizable 3D Object Detection in BEV with single source data, and extensive experiments on various testing domains have demonstrated that our approach achieves the best performance on various test domains with single source data.

Primary Area: applications to computer vision, audio, language, and other modalities

Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.

Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.

Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.

Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.

No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.

Submission Number: 6082