Turn Signal Prediction Dataset

This dataset was created to support research on turn signal detection and prediction using accessible, low-cost hardware.
It contains multimodal driving data collected with a Samsung Galaxy A50 smartphone (inertial sensors: accelerometer and gyroscope) and a dashboard camera recording the road at 1080p and 60 FPS.
The recording session covers approximately 20 minutes of urban driving, resulting in a 2.43 GB video file and a sensor log with 92,778 entries (9.22 MB).

Each sensor entry includes timestamped acceleration and angular velocity data on three axes, together with a Boolean indicator of turn signal use.
All data have been synchronized with the corresponding video frames to allow multimodal analysis of vehicle motion and driver behavior.

本数据集旨在支持基于易用且低成本硬件的转向灯检测与预测相关研究。
其包含多模态驾驶数据，采集设备为三星Galaxy A50智能手机（内置惯性传感器（inertial sensors），包括加速度计（accelerometer）与陀螺仪（gyroscope）），以及一台以1080p分辨率、60帧每秒录制路面的行车记录仪。
本次录制涵盖约20分钟的城市驾驶场景，最终生成2.43 GB的视频文件，以及包含92778条数据条目的传感器日志（文件大小为9.22 MB）。
每条传感器数据条目均包含带时间戳的三轴加速度与角速度数据，同时附带转向灯使用状态的布尔标识。
所有数据均已与对应视频帧完成同步，可用于车辆运动与驾驶员行为的多模态分析。