Human Gait Dataset for Biometric Authentication

The Human Gait Dataset for Biometric Authentication consists of gait patterns collected from 87 individuals across multiple sessions using smartphone motion sensors. Each participant contributed data in two sessions, with three repetitions per session, and each session lasted approximately 1:30 minutes. The dataset was recorded using a self-developed web-based application, capturing a wide range of sensor-based features, including accelerometer, gyroscope, and rotation sensor data, along with demographic details such as gender, age, handedness, and educational qualification. The dataset includes linear acceleration measurements in the x, y, and z axes, rotational movement data from the gyroscope, gravitational force readings, and device orientation angles (Alpha, Beta, Gamma). These features provide valuable insights into gait-based authentication, making this dataset useful for developing biometric authentication systems, machine learning-based anomaly detection models, healthcare applications such as fall detection and rehabilitation analysis, and security applications for continuous authentication. The structured collection of gait data across multiple sessions ensures variability and robustness, allowing researchers to explore authentication strategies that leverage human movement patterns.

用于生物特征认证的人类步态数据集（Human Gait Dataset for Biometric Authentication）收录了87名受试者在多轮采集会话中通过智能手机运动传感器获取的步态模式。每位受试者完成两次数据采集会话，每会话包含三次重复采集流程，单次会话时长约1分30秒。该数据集通过自研的网页端应用完成录制，采集了丰富的传感器特征数据，涵盖加速度计（accelerometer）、陀螺仪（gyroscope）与旋转传感器数据，同时包含受试者的性别、年龄、利手性及学历等人口统计学信息。数据集包含x、y、z三轴的线性加速度测量值、陀螺仪的旋转运动数据、重力读数以及设备姿态角（Alpha、Beta、Gamma）。上述特征为基于步态的认证研究提供了极具价值的参考依据，使得该数据集可用于开发生物特征认证系统、基于机器学习的异常检测模型、跌倒检测与康复分析等医疗应用，以及持续认证类安全应用。多会话场景下的结构化步态数据采集保障了数据集的多样性与鲁棒性，能够支持研究人员探索基于人类运动模式的认证策略。