Dataset for "Understanding Freehand Cursorless Pointing Variability and Its Impact on Selection Performance"

This dataset supports the journal entry "Understanding Freehand Cursorless Pointing Variability and Its Impact on Selection Performance" (TOCHI, 2025), containing motion capture data, of the body and hands, captured during a range of pointing gestures from 23 participants.

The user study that captured this data systematically explored how target position (3 rows by 5 columns), task focus (Pointing as a Primary Task vs Secondary Task), and user effort (Accurate pointing vs Casual pointing), affect pointing behaviour and performance.

The dataset includes:

- Motion capture data for each trial (grouped by participant). This contains body landmarks – captured via a markerless motion capture system and finger landmarks – tracked with infrared markers.
- Trial Annotations. Metadata for each trial, such as the target position, labels for when pointing occurs, and observed behaviour labels.
- Encoded gesture statistics. For each trial, for which a valid pointing gesture could be extracted, an encoding of the gesture performed, derived from the medians for body pose features (e.g. elbow flexion), fatigue measures (e.g. consumed endurance), and rays (e.g. vector and accuracy).
- Self-reported user data. Including participant age, hand dominance, and fatigue measures (obtained after completing pointing within each condition).
- Code for visualising the trials, including a subset of the rays used in our subsequent analysis, code for generating our encoded gestures, using the motion capture data and annotations, and the script used to perform the analysis over our encoded gestures.

This dataset has been provided for two purposes:

1. For further investigation into pointing behaviour and for the development of pointing interaction systems. For this, please refer to the Pointing Dataset section of the README to understand the structure and dataset contents, and the Trial Visualiser section of the README for usage of a script for visualising the motion capture data.

2. For reproduction of data used in the analysis of the accompanying paper (Understanding Freehand Cursorless Pointing Variability and Its Impact on Selection Performance), for which please see Pointing Dataset section of the README to understand the structure and dataset contents, along with the Gesture Encoder and Analysis Script sections of the README for the code used to perform our analysis.

本数据集支持期刊论文《Understanding Freehand Cursorless Pointing Variability and Its Impact on Selection Performance》（TOCHI, 2025），包含23名参与者在一系列指向手势过程中采集的身体与手部动作捕捉数据（motion capture data）。

本次数据采集所依托的用户研究系统探究了目标位置（3行×5列布局）、任务优先级（指向作为主任务 vs 副任务）以及用户操作投入度（精准指向 vs 随意指向）对指向行为与性能的影响。

本数据集包含以下内容：
- 按参与者分组的试次动作捕捉数据：包含通过无标记点动作捕捉系统（markerless motion capture system）采集的身体关键点数据，以及借助红外标记物（infrared markers）追踪的手指关键点数据。
- 试次标注信息：每项试次的元数据，例如目标位置、指向动作发生时刻的标签以及观测行为标签。
- 编码手势统计特征：针对可提取有效指向手势的每项试次，对执行手势进行编码，该编码源自身体姿态特征（如肘关节屈曲角度）、疲劳度量（如消耗的耐力水平）以及指向射线（如方向向量与精度）的中位数。
- 自我报告的用户数据：包括参与者年龄、利手情况，以及在每种实验条件下完成指向任务后采集的疲劳度量数据。
- 试次可视化代码：包含用于可视化部分指向射线数据（用于后续分析）的代码、基于动作捕捉数据与标注信息生成编码手势的代码，以及用于对编码手势开展分析的脚本。

本数据集提供两个使用场景：
1. 用于指向行为的进一步研究以及指向交互系统的开发。如需开展此类工作，请参阅README文件中的「指向数据集」章节以了解数据集结构与内容，并参考README的「试次可视化工具」章节以掌握动作捕捉数据可视化脚本的使用方法。
2. 用于复现配套论文《Understanding Freehand Cursorless Pointing Variability and Its Impact on Selection Performance》分析中所用的数据。相关操作请参阅README的「指向数据集」章节了解数据集结构与内容，并结合README的「手势编码器与分析脚本」章节使用对应分析代码。