StandardSim

StandardSim是由石溪大学和Standard Cognition合作创建的大规模合成数据集，专注于零售环境下的视觉和感知任务。该数据集包含2,134个独特场景，总计25,436张图像，支持多种计算机视觉任务，如语义分割、实例分割、深度估计和对象检测。特别引入了名为‘变化检测’的新任务，旨在模拟零售环境中顾客与商品的互动。数据集通过高度精确的商店模型和详细的标注，为自主结账系统提供了丰富的训练和测试资源，特别适用于解决零售场景中的复杂视觉问题。

StandardSim is a large-scale synthetic dataset co-created by Stony Brook University and Standard Cognition, focusing on visual and perceptual tasks in retail environments. This dataset contains 2,134 unique scenes with a total of 25,436 images, supporting a variety of computer vision tasks including semantic segmentation, instance segmentation, depth estimation, and object detection. Specifically, a novel task named "change detection" was introduced to simulate customer-merchandise interactions in retail scenarios. Leveraging highly precise store models and detailed annotations, the dataset provides rich training and testing resources for autonomous checkout systems, and is particularly suitable for addressing complex visual problems in retail scenarios.