自动驾驶数据集

<p>用于自动驾驶空间理解的数据集，通过激光雷达语义分割模型，实现自动驾驶中稳健可靠的 3D 语义分割，从而构建一个全面的激光雷达语义分割稳健性基准。它包含 16 种域外激光雷达损坏情况，分为三组，即恶劣天气、测量噪声和跨设备差异。SemanticKITTI-C自动驾驶数据集</p><p>数据来源：公开网站<a href="https://opendatalab.org.cn/OpenDataLab/SemanticKITTI-C" rel="noopener noreferrer" target="_blank">https://opendatalab.org.cn/OpenDataLab/SemanticKITTI-C</a></p><p>数据规模：近一百个点云场景</p><p>数据特点：覆盖360度的场景扫描</p><p>应用场景：自动驾驶汽车训练</p>This dataset for spatial understanding in autonomous driving leverages LiDAR semantic segmentation models to achieve robust and reliable 3D semantic segmentation in autonomous driving scenarios, thereby establishing a comprehensive robustness benchmark for LiDAR semantic segmentation. The SemanticKITTI-C autonomous driving dataset contains 16 types of out-of-distribution (OOD) LiDAR corruptions, divided into three groups: adverse weather, measurement noise, and cross-device discrepancies. Data Source: Publicly available at <a href="https://opendatalab.org.cn/OpenDataLab/SemanticKITTI-C" rel="noopener noreferrer" target="_blank">https://opendatalab.org.cn/OpenDataLab/SemanticKITTI-C</a> Data Scale: Nearly one hundred point cloud scenes Data Characteristics: Covers 360-degree scene scanning Application Scenarios: Autonomous vehicle training