Abstract
Recent advances in deep learning have significantly improved 3D semantic segmentation, but most models focus on indoor or terrestrial datasets. Their behavior under real aerial acquisition conditions remains insufficiently explored, and although a few studies have addressed similar scenarios, they differ in dataset design, acquisition conditions, and model selection. To address this gap, we conduct an experimental benchmark evaluating several state-of-the-art architectures on a large-scale aerial LiDAR dataset acquired under operational flight conditions in Navarre, Spain, covering heterogeneous urban, rural, and industrial landscapes. This study compares four representative deep learning models, including KPConv, RandLA-Net, Superpoint Transformer, and Point Transformer V3, across five semantic classes commonly found in airborne surveys, such as ground, vegetation, buildings, and vehicles, highlighting the inherent challenges of class imbalance and geometric variability in aerial data. Results show that all tested models achieve high overall accuracy exceeding 93%, with KPConv attaining the highest mean IoU (78.51%) through consistent performance across classes, particularly on challenging and underrepresented categories. Point Transformer V3 demonstrates superior performance on the underrepresented vehicle class (75.11% IoU), while Superpoint Transformer and RandLA-Net trade off segmentation robustness for computational efficiency.
Abstract (translated)
深度学习的最新进展显著提升了3D语义分割性能，但多数模型聚焦于室内或地面数据集。其在真实航拍采集条件下的表现尚未得到充分探索，尽管少数研究涉及类似场景，但在数据集设计、采集条件和模型选择上存在差异。为填补这一空白，本研究通过实验基准测试，在西班牙纳瓦拉地区 operational flight 条件下获取的大规模航拍激光雷达数据集上，评估了多个先进架构。该数据集覆盖异质性城市、乡村及工业景观。研究对比了四种代表性深度学习模型（包括KPConv、RandLA-Net、Superpoint Transformer和Point Transformer V3），针对航测中常见的五类语义目标（如地面、植被、建筑物和车辆）进行分析，凸显了航拍数据中固有的类别不平衡与几何变异挑战。结果表明，所有测试模型整体准确率均超过93%，其中KPConv通过跨类别的稳定表现达到最高平均交并比（78.51%），尤其在具有挑战性的低占比类别上表现突出。Point Transformer V3在低占比的车辆类别上展现最优性能（交并比75.11%），而Superpoint Transformer与RandLA-Net则在分割鲁棒性与计算效率间取得权衡。
URL
https://arxiv.org/abs/2603.22229
