UWB Trajectories and Fine-grained Stop-Move Detection: A Museum Dataset

This collection of datasets captures the visit experience of  a few museum visitors tracked using a UWB localization system. The datasets include: UWB trajectories,  the museum space layout, and  the semantic trajectories derived from segmenting UWB trajectories using various stop-detection methods. Data were collected in the same area at two different times and under two distinct setups, referred to as in-vitro and in-vivo. The in-vitro setup enabled a fully controlled experiment where visitors followed constrained movement patterns, while the in-vivo setup allowed visitors to move freely within the study area.

Datasets are grouped in two directories:

In Vitro

In Vivo

Each directory is composed of:

Original trajectories: Contains a file encompassing all the original WB trajectory data, along with the associated fields required for visualization, analysis, and re-generation of the experiments and results presented in paper [1].

Ground truth: The true stops are reported.

SeqScan output, SPD output, KBV output: Each includes the output of the corresponding trajectory segmentation method (SeqScan, SPD, and KBV). These results are reported both as points and as stops, with parameters corresponding to those detailed in Tables 4, 6, 7 and 8 of paper [1]. Furthermore, the in-vivo directory also includes visits data, with parameters reported in Tables 9 and 10 of paper [1].

Museum objects: Includes the coordinates of the exhibits (POIs) inside the museum. As points in case of in-vitro and as segments in case of in-vivo.

For a detailed description, please refer to the README document.

[1] Fatima Hachem, Davide Vecchia, Maria Luisa Damiani, Gian Pietro Picco. "Fine-grained Stop-Move Detection with UWB: Quality Metrics and Real-world Evaluation". Under review at ACM Transactions on Sensor Networks, 2025.

本数据集集合收录了利用超宽带（UWB）定位系统追踪的若干博物馆参观者的参观体验相关数据。数据集包含以下内容：UWB轨迹、博物馆空间布局，以及通过多种停留检测算法对UWB轨迹进行分段后得到的语义轨迹。

数据于同一区域内的两个不同时段、以两种差异化设置采集完成，分别称为in-vitro与in-vivo场景。其中in-vitro场景可开展完全受控的实验，参观者需遵循预设的受限移动模式；而in-vivo场景允许参观者在研究区域内自由移动。

数据集分为两个目录：
In Vitro
In Vivo

每个目录均包含以下内容：
原始轨迹：包含涵盖全部原始UWB轨迹数据的文件，以及用于可视化、分析、复现论文[1]中所述实验与结果所需的关联字段。
真值标签：记录了真实的停留点信息。
SeqScan输出、SPD输出与KBV输出：分别对应三种轨迹分段算法（SeqScan、SPD、KBV）的运行结果。上述结果均以轨迹点与停留点两种形式呈现，其参数设置与论文[1]中表4、6、7、8所列参数一致。此外，in-vivo目录还包含参观数据，其参数设置对应论文[1]中表9与表10的内容。
博物馆展品信息：包含博物馆内展品（POI，Point of Interest）的坐标信息。其中in-vitro场景下展品坐标以点形式存储，in-vivo场景下则以线段形式存储。

如需了解详细说明，请参阅README文档。

[1] Fatima Hachem、Davide Vecchia、Maria Luisa Damiani、Gian Pietro Picco：《基于UWB的细粒度停留-移动检测：质量指标与真实场景评估》，提交至ACM Transactions on Sensor Networks，2025年。