Reference and Target Datasets with the Point-Descriptor-Precedence representation

There are four reference datasets generated using the animations given in slides 1-4 of the Animation.pptx: Reference Movement Pattern with one-point, two-points, four-points, and two-points for the length, width and speed analysis. The animations are stored in the video format and then a sequence of images were extracted to create static reference datasets/movement patterns. The Target dataset consists of 46 vehicles and is created using the animation given in Target.pptm, which has a built in macro to run the simulation and save it in a video format. The video is then used to extract images which matched exactly with the reference datasets. This resulted in finding the exact matches of the reference datasets. The code is written in Python and can be run with any Python IDE.本数据集包含4个参考数据集，均基于Animation.pptx第1至4页的动画生成：分别为适配长度、宽度与速度分析的单点、两点、四点及两点参考运动模式（Reference Movement Pattern）。上述动画均以视频格式存储，随后通过提取连续图像帧以构建静态参考数据集/运动模式。 目标数据集包含46个车辆对象，基于Target.pptm内的动画生成；该文件内置宏指令以运行仿真并将仿真结果保存为视频格式。随后从该视频中提取与参考数据集完全匹配的图像序列，最终成功获取了与参考数据集精准匹配的样本。 本次实验所用代码基于Python编写，可在任意Python集成开发环境（IDE）中运行。