一、北京师范大学大型自发视觉情感数据库2.0版(BNU-LSVED2.0)是第一个课堂环境下的大规模自发多模态学生情感数据库。

详情可见:
Wei Q, Bo S, He J, et al. BNU-LSVED 2.0: Spontaneous multimodal student affect database with multi-dimensional labels[J]. Signal Processing Image Communication, 2017:S0923596517301510.

北京师范大学静态维度情感数据库(BNU-SDED)是由BNU-LSVED 2.0中的关键帧图片挑选形成。

如何获取数据集?
如需获取数据集,请发送电子邮件至hejun@bnu.edu.cn,主题为:“BNU-LSVED download”,并上传带有签名的许可证文件[许可证PDF]。

The Beijing Normal University Large-scale Spontaneous Visual Expression Database version 2.0 (BNU-LSVED2.0 ) is the first large-scale spontaneous and multimodal student affect database in the classroom environment. The details can be found in:

Wei Q, Bo S, He J, et al. BNU-LSVED 2.0: Spontaneous multimodal student affect database with multi-dimensional labels[J]. Signal Processing Image Communication, 2017:S0923596517301510.

The Beijing Normal University Static Dimensional Expression Database (BNU-SDED ) has been developed by selecting key frames from BNU-LSVED 2.0.
How to obtain access to the images:

To obtain a copy, please email at hejun@bnu.edu.cn with the subject: “BNU-LSVED download” with signed licence document [Licence PDF].


二、北京师范大学学生课堂行为数据集(BNU-LCSAD)

我们收集来自11间教室、不同学科17门课程的128节(40分钟左右)的视频,构建了基于真实课堂的北京师范大学大规模学生课堂、课间自发行为数据集(BNU-LCSAD、SBBRD)。学生课堂行为数据集提供3类标签,包括行为检测、识别和视频描述,包括时间检测数据模块(4542个样本)、行为检测数据模块(3343个样本)、行为识别模块(4276个样本)和视频描述模块(4296个样本);与此同时,我们挑选64段课间视频,构建了包含9类行为的1238条样本的学生课间行为数据集。

针对课堂行为数据集,2021年发表在“Neural Computing and Applications”上的文章“Student Class Behavior Database: a video dataset for recognizing, detecting, and captioning students’ behaviors in classroom scenes”给出了详细介绍,分析了课堂场景的特点和各模块(任务)的技术难点,并通过实验验证了数据集的有效性和可信性。由于教室的特殊性,如遮挡、多被试、被试规模差异大、动作类别差异小等,数据集对现有方法提出了更高的要求。此外,我们为数据集中的每个任务模块提供了一个baseline,并与当前主流数据集进行了比较。结果表明,我们的数据集是可行和可靠的。代码可在线下载:https: //github.com/BNU-Wu/Student-Class-Behavior-Dataset/tree/master。

课间行为数据集基于北京师范大学的64段视频。经过处理,得到1538个包含9个学生课间行为的视频样本;我们为该数据集也提供了一个baseline。论文“Student break behavior recognition dataset”发布在2021 6th International Conference on Image, Vision and Computing (ICIVC)。

由于涉及到隐私问题,我们只能提供提取好的特征信息,敬请谅解!目前识别模块数据已经处理完,其他的还在处理。

如何获取数据集?
如需获取数据集,请发送电子邮件至tosunbo@bnu.edu.cn,hejun@bnu.edu.cn,主题为:“BNU-LCSAD download”,并上传带有签名的许可证文件agreement for LCSAD


三、北京师范大学学生交互行为视觉数据集(SVIBD)

we constructed the Beijing Normal University Student Visual Interactive Behaviour Dataset (BNUSVIBD), which was collected from 10 daily instructional videos
in a classroom at Beijing Normal University. After processing, 114 interactive sequence samples, 332 interactive subsequence samples and 4622 common classroom interaction behaviours were obtained.

四、北京师范大学学生启发式教育数据集(BNU-MHED)

Heuristic teaching is an active teaching pattern which emphasizes elicitation. Corresponding high-quality samples are primal for developing Artificial Intelligence technology for evaluating heuristic. Thus, we present a BNU-MHED, i.e. a Multi-modal Heuristic Education Dataset from Beijing Normal University. It includes 3 tasks, namely heuristic teaching action detection, recognition and classification.

如何获取数据集?
如需获取数据集,请发送电子邮件至tosunbo@bnu.edu.cn,hejun@bnu.edu.cn,主题为:“BNU-MHED download”,并上传带有签名的许可证文件agreement for MHED