Event
SynCalib Improves LiDAR-Camera Calibration Generalization with Extrinsically Controllable Synthetic Data
1 reports · 1 sources
Huihua Zhu et al. propose the SynCalib data generation framework to address overfitting and poor generalization caused by insufficient extrinsic diversity in training data for deep-learning-based online LiDAR-camera calibration. Using panoramic image datasets with depth ground-truth annotations, it synthesizes multimodal calibration data under strict extrinsic constraints. The paper reports that experiments show this pipeline significantly improves the generalization and robustness of calibration networks, and real-world evaluations were completed across multiple camera-LiDAR configurations, verifying that models trained solely on synthetic data possess zero-shot transfer capability; the data generation code, pretrained models, and online benchmark dataset will be made publicly available.
Report timeline
- SynCalib: Breaking Dataset Shortcuts in LiDAR-Camera Calibration via Extrinsic-Controllable Synthesis
The paper proposes SynCalib, a data generation framework that addresses overfitting and poor generalization caused by insufficient extrinsic diversity in training data for deep learning-based online LiDAR-camera calibration. Using omnidirectional image datasets annotated with depth ground truth, it synthesizes multimodal calibration data under strict extrinsic constraints. Experiments show that this pipeline significantly improves the generalization and robustness of learning-based calibration networks, and real-world evaluations across various camera-LiDAR configurations validate the zero-shot transfer capability of models trained solely on synthetic data; the data generation code, pretrained models, and online benchmark dataset will be made public.
arXiv cs.RO ·