合成数据集生成工具

本数据集由英特尔公司开发，旨在通过模拟环境生成大规模的合成数据集，以研究对抗机器学习。数据集利用CARLA模拟器收集，包含多模态感测数据、实例分割标签、多样化的背景和天气条件，以及多视角场景视图。该数据集特别关注于创建真实的对抗性示例，这些示例经历了与现实世界图像相同的处理和环境转换。数据集的应用领域主要集中在评估对抗性示例的有效性，以及提高机器学习模型在安全关键应用中的鲁棒性。

This dataset was developed by Intel Corporation, with the objective of generating large-scale synthetic datasets through simulated environments for adversarial machine learning research. The dataset is collected using the CARLA simulator, and includes multimodal sensory data, instance segmentation labels, diverse background and weather conditions, as well as multi-view scene views. This dataset specifically focuses on creating realistic adversarial examples that undergo the same processing and environmental transformations as real-world images. The primary application scenarios of this dataset are centered on evaluating the effectiveness of adversarial examples, and enhancing the robustness of machine learning models in safety-critical applications.