Students Attention Detection Dataset
收藏DataCite Commons2025-05-01 更新2025-05-17 收录
下载链接:
https://data.mendeley.com/datasets/smzggbnkd2
下载链接
链接失效反馈官方服务:
资源简介:
The purpose of the creation of this dataset for analyzing student's attention and behavior based on various visual cues. This dataset is generated using a combination of fundamental modules that enable the extraction of distinct high-level features from video frames. The key components involved in the dataset creation process are face detection, hand tracking, mobile phone detection, and pose estimation. These components collectively produce a set of features that characterize different aspects of student presence and orientation within the observed frames.
The face detection module yields the count of detected faces, along with their coordinates, width, and height. The hand tracking module provides the count of identified hands. The pose estimation component generates head orientation, x-axis rotation, and y-axis rotation. The mobile phone detection module outputs six features: mobile phone presence, phone coordinates, width, height, and detection confidence.
The dataset comprises a total of 17 columns, which consist of 16 feature columns and one label column. In total, there are 4,000 records within the dataset. The features are described as follows:
no_of_face: Indicates the number of faces detected in a frame.
face_x: Represents the x-coordinate of the upper-left corner of a detected face.
face_y: Denotes the y-coordinate of the upper-left corner of a detected face.
face_w: Reflects the width of a detected face in pixels.
face_h: Specifies the height of a detected face in pixels.
face_con: Provides the confidence score for face detection.
no_of_hands: Indicates the number of hands detected in a frame.
pose: Describes the orientation of a detected face (Forward, Left, Right, Down).
pose_x: Represents the x-axis rotation of the detected face's orientation.
pose_y: Represents the y-axis rotation of the detected face's orientation.
phone: Indicates the presence of a mobile phone (0: No phone detected, 1: Phone detected).
phone_x: Specifies the x-coordinate of the upper-left corner of a detected phone.
phone_y: Denotes the y-coordinate of the upper-left corner of a detected phone.
phone_w: Reflects the width of a detected phone in pixels.
phone_h: Specifies the height of a detected phone in pixels.
phone_con: Provides the confidence score for phone detection.
label: Serves as the target column, indicating whether the subject is attentive (0) or inattentive (1).
The features encompass a wide range of factors that contribute to assessing a student's level of attentiveness. The diverse set of features and corresponding label column of the dataset make it a valuable resource for training machine learning models aimed at recognizing and classifying attentive and inattentive behaviors of the student.
本数据集旨在基于多种视觉线索分析学生的注意力状态与行为表现。该数据集通过整合基础模块生成,这些模块可从视频帧中提取独特的高层特征。数据集构建过程涉及的核心组件包括人脸检测(face detection)、手部追踪(hand tracking)、手机检测(mobile phone detection)与姿态估计(pose estimation)。上述组件协同生成一系列特征,用以刻画观测帧中学生的在场状态与朝向姿态等不同维度的信息。
人脸检测模块可输出检测到的人脸数量,以及人脸的坐标、宽度与高度。手部追踪模块提供识别到的手部数量。姿态估计组件可生成头部朝向、x轴旋转角与y轴旋转角。手机检测模块输出六项特征:手机存在状态、手机坐标、宽度、高度与检测置信度。
本数据集共计包含17列,其中16列为特征列,1列为标签列,总共有4000条数据记录。各特征说明如下:
人脸数量(no_of_face):表示单帧图像中检测到的人脸总数。
人脸x坐标(face_x):表示检测到的人脸左上角的x轴坐标。
人脸y坐标(face_y):表示检测到的人脸左上角的y轴坐标。
人脸宽度(face_w):以像素为单位,表示检测到的人脸的宽度。
人脸高度(face_h):以像素为单位,表示检测到的人脸的高度。
人脸检测置信度(face_con):提供人脸检测的置信评分。
手部数量(no_of_hands):表示单帧图像中检测到的手部总数。
头部朝向(pose):描述检测到的人脸的朝向(正向、左向、右向、向下)。
x轴旋转角(pose_x):表示检测到的人脸朝向的x轴旋转角度。
y轴旋转角(pose_y):表示检测到的人脸朝向的y轴旋转角度。
手机存在状态(phone):表示是否检测到手机,其中0代表未检测到手机,1代表检测到手机。
手机x坐标(phone_x):表示检测到的手机左上角的x轴坐标。
手机y坐标(phone_y):表示检测到的手机左上角的y轴坐标。
手机宽度(phone_w):以像素为单位,表示检测到的手机的宽度。
手机高度(phone_h):以像素为单位,表示检测到的手机的高度。
手机检测置信度(phone_con):提供手机检测的置信评分。
标签(label):作为目标列,用于标注受试者是否处于专注状态,其中0代表专注,1代表分心。
上述特征涵盖了评估学生注意力水平所需的多维度影响因素。本数据集丰富的特征集合与配套的标签列,使其成为训练机器学习模型的优质资源,可用于识别并分类学生的专注与分心行为。
提供机构:
Mendeley Data
创建时间:
2023-08-08
搜集汇总
数据集介绍

背景与挑战
背景概述
该数据集是一个用于分析学生注意力和行为的数据集,包含4000条记录和17个特征列,基于面部检测、手部跟踪、手机检测和姿势估计等多种视觉线索。数据集适用于训练机器学习模型,以识别和分类学生的专注和非专注行为。
以上内容由遇见数据集搜集并总结生成



