多视角多模态驾驶员综合分析数据集

多视角多模态驾驶员综合分析数据集主要包含以下三部分内容，合计18.2G：1. 驾驶员状态数据：采集驾驶员车内驾驶状态数据，估计车内驾驶员疲劳驾驶等状态；2.驾驶员行为数据：采集驾驶员车内行为动作数据，识别车内驾驶员分心驾驶等行为；3.驾驶员意图数据：采集车内驾驶员不同意图数据，预测驾驶员转变变道驾驶等意图。
自行采集方案：1.状态数据：车内一个摄像头，采集驾驶员在正常驾驶、打哈欠、说话、疲劳点头、向外张望五类驾驶状态数据。2.行为数据：车内一个摄像头，采集驾驶员安全驾驶、左右手看手机、左右手接打电话、调音乐、喝水、整理头发、与乘客交谈、向后取东西等十类分心行为数据。3.意图数据：两个摄像头分别采集驾驶员直行、左转弯、右转弯、左变道、右变道五类意图数据；车外一个摄像头，采集车外直行、左转弯、右转弯、左变道、右变道时相应五类意图数据。

This multi-view and multi-modal comprehensive driver analysis dataset consists of the following three parts, with a total size of 18.2 GB:
1. Driver State Data: Collect in-vehicle driving status data of drivers, to estimate driving states including fatigued driving.
2. Driver Behavior Data: Collect in-vehicle driving action data of drivers, to identify distracted driving behaviors such as in-vehicle driver distraction.
3. Driver Intention Data: Collect various intention data of in-vehicle drivers, to predict driving intentions such as lane changes.

Self-developed data collection scheme:
1. Driver State Data: One in-vehicle camera is deployed to collect five types of driving state data, including normal driving, yawning, speaking, fatigued nodding, and looking outward.
2. Driver Behavior Data: One in-vehicle camera is deployed to collect ten types of distracted driving behavior data, including safe driving, using mobile phones with left/right hands, making phone calls with left/right hands, adjusting music, drinking water, tidying hair, talking with passengers, and retrieving items from the rear.
3. Driver Intention Data: Two in-vehicle cameras are deployed to collect five types of driving intention data, including driving straight, left turn, right turn, left lane change, and right lane change; one exterior vehicle camera is deployed to collect corresponding five types of intention data when the driver conducts straight driving, left turn, right turn, left lane change, and right lane change.

搜集汇总

背景与挑战

背景概述

该数据集是一个多视角多模态的驾驶员综合分析数据集，总数据量18.28GB，包含驾驶员状态、行为和意图三部分数据，用于估计疲劳驾驶、识别分心行为和预测变道意图等。数据通过车内和车外摄像头采集，覆盖多种驾驶场景，如正常驾驶、打哈欠、看手机、转弯等，支持人机共驾系统的研究与应用。