LiDAR has been widely used in autonomous driving and robotic navigation, while ground segmentation is a key preprocessing step. However, existing methods often suffer from over segmentation, under segmentation or insufficient real-time performance. In this letter, RACG-Ground, Ring-Adaptive Confidence Gating is developed for LiDAR Ground Segmentation, which is a fast and purely geometry-based ground segmentation approach. The method greedily aggregates fixed step radial cells into adaptive polar cells, which improves radial uniformity and reduces empty cells under range-dependent sparsity. On this grid, a cell-level confidence gating mechanism is designed in which vertical dispersion, upward offset, and lateral gradient are fused, with frame-wise adaptive thresholds employed to handle scene changes. A robust radial ground filter is then performed along each sector with lateral consensus, followed by point-wise multi- criterion gating that combines asymmetric tolerance bands and reference-height consistency for the final classification. Extensive experiments on SemanticKITTI (64 beams), nuScenes (32 beams), and SemanticTHAB (128 beams), together with real-world field tests show that RACG-Ground is accurate and real-time. On SemanticKITTI, it achieves 94.34% F1 at an average of 187 Hz, indicating robustness to outliers and occlusion-induced shadows.
Yu-Qi Shi, Weiwei Lyu, Shuanggen Jin et al.· IEEE Robotics and Automation...· 0 citations
Scene graph generation (SGG) addresses the task of detecting objects in an image and predicting the relationships among them. Although prototype-based methods have recently achieved clear progress on long-tailed SGG, fine-grained low-frequency predicates remain difficult to recognize because their relation features often exhibit larger intra-class variation and more dispersed distributions, making them easily confused with semantically similar high-frequency coarse-grained predicates under a unified prototype-matching rule. To alleviate this issue, we propose a frequency-aware elastic prototype boundary learning framework, termed SGE-Net. Under fixed relation prototypes, the framework learns relation-category-specific boundary scales through explicit frequency compensation and frequency-adaptive virtual sampling, so that relation prediction can exploit not only prototype-center matching but also category-dependent decision-boundary information. During inference, we further introduce elastic boundary-aware distance calibration, enabling the boundary information learned during training to better distinguish relation categories that are easily confused under prototype matching. In addition, we combine visual and semantic features with dynamic gating to provide more reliable relation features for the above boundary learning. Experiments and analyses on Visual Genome and Open Images V6 demonstrate that the proposed method achieves consistent gains in both long-tailed relation prediction and overall evaluation metrics.
Binghao Wang, Xueying Sun, Hanzhu Dai et al.· Journal of King Saud Univers...· 0 citations
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
