Data Underlying the Publication: Studying the Effect of Explicit Interaction Representations on Learning Scene-level Distributions of Human Trajectories

This dataset contains four folders with results on the four naturalistic datasets used in the publication (Argoverse, INTERACTION, NuScenes, and RounD).These four folders each containing the processed data (Data folder), dataset splitting (Splitting), trained models (Models), predictions obtained from the trained models (Predictions), and the final calculated metrics (Metrics) This data is to be used in conjunction to the code available at https://github.com/anna-meszaros/GMoP-Experiment-Setup together with https://github.com/DAI-Lab-HERALD/General-Framework, and contains the results obtained using this code. The data from the zip file, should be removed from the '[DATASET]_Results' folder and placed in the Results folder within the General-Framework.

本数据集包含四个文件夹，分别对应论文中所采用的四个自然场景数据集（Argoverse、INTERACTION、NuScenes与RounD）的实验结果。每个子文件夹均涵盖预处理后的数据（Data文件夹）、数据集划分方案（Splitting）、已训练模型（Models）、基于已训练模型生成的预测结果（Predictions），以及最终计算得到的评估指标（Metrics）。本数据集需与公开代码库https://github.com/anna-meszaros/GMoP-Experiment-Setup及https://github.com/DAI-Lab-HERALD/General-Framework协同使用，其内存储的即为通过上述代码得到的实验结果。请将压缩包中的数据从「[DATASET]_Results」文件夹中移出，并放置于General-Framework项目内的Results文件夹中。