PSI(IUPUI-CSRC Pedestrian Situated Intent)

IUPUI-CSRC 行人定位意图 (PSI) 基准数据集除了全面的计算机视觉注释外，还有两个创新标签。第一个新颖的标签是行人在自驾车前过马路的动态意图变化，由 24 位不同背景的司机实现。第二个是在估计行人意图并预测他们在交互期间的行为时对驾驶员推理过程的基于文本的解释。

IUPUI-CSRC Pedestrian Situational Intent (PSI) Benchmark Dataset includes two innovative labels in addition to comprehensive computer vision annotations. The first novel label covers dynamic changes in pedestrians' crossing intent when they walk across the road in front of autonomous vehicles, which was annotated by 24 drivers with diverse backgrounds. The second label consists of text-based explanations of drivers' reasoning processes when estimating pedestrian intent and predicting their behaviors during human-vehicle interactions.

搜集汇总

背景与挑战

背景概述

PSI数据集是一个行人定位意图基准数据集，包含行人在自驾车前过马路的动态意图变化和驾驶员推理过程的文本解释，由多所大学和丰田北美公司于2021年发布。