Bus Violence: a large-scale benchmark for video violence detection in public transport

<strong>Dataset</strong> The <em>Bus Violence </em>dataset<em> </em>is a large-scale collection of videos depicting violent and non-violent situations in public transport environments. This benchmark was gathered from multiple cameras located inside a moving bus where several people simulated violent actions, such as stealing an object from another person, fighting between passengers, etc. It contains 1,400 video clips manually annotated as having or not violent scenes, making it one of the biggest benchmarks for video violence detection in the literature. Specifically, videos are recorded from three cameras at 25 Frames Per Second (FPS) --- two cameras located in the corners of the bus (with resolution 960x540 px) and one fisheye in the middle (1280x960 px). The clips have a minimum length of 16 frames and a maximum of 48 frames, capturing a very precise action (either violence or non-violence). The dataset is perfectly balanced, containing 700 videos of violence and 700 videos of non-violence. The <em>Bus Violence</em> dataset is intended as a test data benchmark. However, for researchers interested in using our data also for training purposes, we provide training and test splits. In this repository, we provide the 1,400 video clips divided into two folders named Violence /NoViolence, containing clips of violent situations and non-violent situations, respectively; two txt files containing the names of the videos belonging to the training and test splits, respectively. <strong>Citing our work</strong> If you found this dataset useful, please cite the following paper <pre>@inproceedings{bus_violence_dataset_2022, title = {Bus Violence: An Open Benchmark for Video Violence Detection on Public Transport}, doi = {10.3390/s22218345}, url = {https://doi.org/10.3390%2Fs22218345}, year = 2022, month = {oct}, publisher = {{MDPI} {AG}}, volume = {22}, number = {21}, pages = {8345}, author = {Luca Ciampi and Pawe{\l} Foszner and Nicola Messina and Micha{\l} Staniszewski and Claudio Gennaro and Fabrizio Falchi and Gianluca Serao and Micha{\l} Cogiel and Dominik Golba and Agnieszka Szcz{\k{e}}sna and Giuseppe Amato}, journal = {Sensors} } </pre> and this Zenodo Dataset <pre>@dataset{pawel_bus_violence_zenodo, author = {Paweł Foszner, Michał Staniszewski, Agnieszka Szczęsna, Michał Cogiel, Dominik Golba, Luca Ciampi, Nicola Messina, Claudio Gennaro, Fabrizio Falchi, Giuseppe Amato, Gianluca Serao}, title = {{Bus Violence: a large-scale benchmark for video violence detection in public transport}}, month = sep, year = 2022, publisher = {Zenodo}, version = {1.0.0}, doi = {10.5281/zenodo.7044203}, url = {https://doi.org/10.5281/zenodo.7044203} } </pre> <strong>Contact Information</strong> Blees Sp. z o.o., Gliwice, Poland<br> [e-mail] <strong>Acknowledgments</strong> The presented dataset was supported by: European Union funds awarded to Blees Sp. z o.o. under grant POIR.01.01.01-00-0952/20-00 “Development of a system for analysing vision data captured by public transport vehicles interior monitoring, aimed at detecting undesirable situations/behaviours and passenger counting (including their classification by age group) and the objects they carry”); EC H2020 project "AI4media: a Centre of Excellence delivering next generation AI Research and Training at the service of Media, Society and Democracy" under GA 951911; research project INAROS (INtelligenza ARtificiale per il mOnitoraggio e Supporto agli anziani), Tuscany POR FSE CUP B53D21008060008. <strong>License</strong> The <em>Bus Violence </em>dataset was acquired by Blees Sp. z o.o. and is released under a Creative Commons Attribution license for non-commercial use.<strong>数据集</strong> <em>公交暴力（Bus Violence）</em>数据集是一款大规模收录公共交通场景下暴力与非暴力行为的视频数据集。该基准数据集采集自运行中公交车内的多台摄像头，由多名受试者模拟各类暴力行为，例如从他人处窃取物品、乘客间斗殴等。数据集共包含1400段经人工标注的视频片段，明确区分含暴力场景与不含暴力场景的样本，是当前学术文献中规模最大的视频暴力检测（video violence detection）基准数据集之一。具体而言，视频由三台摄像头以25帧每秒（Frames Per Second, FPS）的帧率录制：两台布置于公交车车厢角落，分辨率为960×540像素；另一台为鱼眼（fisheye）摄像头，安装于车厢中部，分辨率为1280×960像素。所有视频片段的长度介于16帧至48帧之间，精准捕捉单类行为（暴力或非暴力）。该数据集样本分布完全均衡，含暴力与非暴力视频各700段。<em>公交暴力（Bus Violence）</em>数据集最初设计为测试基准数据集，但针对希望将其用于模型训练的研究人员，我们同时提供了训练集与测试集的划分方案。本仓库中，我们将1400段视频按类别分为两个文件夹：Violence与NoViolence，分别存放暴力场景与非暴力场景的视频片段；同时提供两个文本文件，分别记录训练集与测试集的视频文件名。 <strong>引用本数据集</strong> 若您认为本数据集对您的研究有所帮助，请引用以下论文及Zenodo数据集： <pre>@inproceedings{bus_violence_dataset_2022, title = {Bus Violence: An Open Benchmark for Video Violence Detection on Public Transport}, doi = {10.3390/s22218345}, url = {https://doi.org/10.3390%2Fs22218345}, year = 2022, month = {oct}, publisher = {{MDPI} {AG}}, volume = {22}, number = {21}, pages = {8345}, author = {Luca Ciampi and Pawe{l} Foszner and Nicola Messina and Micha{l} Staniszewski and Claudio Gennaro and Fabrizio Falchi and Gianluca Serao and Micha{l} Cogiel and Dominik Golba and Agnieszka Szcz{k{e}}sna and Giuseppe Amato}, journal = {Sensors} } </pre> <pre>@dataset{pawel_bus_violence_zenodo, author = {Pawe{l} Foszner, Michał Staniszewski, Agnieszka Szczęsna, Michał Cogiel, Dominik Golba, Luca Ciampi, Nicola Messina, Claudio Gennaro, Fabrizio Falchi, Giuseppe Amato, Gianluca Serao}, title = {{Bus Violence: a large-scale benchmark for video violence detection in public transport}}, month = sep, year = 2022, publisher = {Zenodo}, version = {1.0.0}, doi = {10.5281/zenodo.7044203}, url = {https://doi.org/10.5281/zenodo.7044203} } </pre> <strong>联系方式</strong> 波兰格利维采的Blees Sp. z o.o.；邮箱：[e-mail] <strong>致谢</strong> 本数据集的研发得到以下项目资助：欧盟为Blees Sp. z o.o.拨付的POIR.01.01.01-00-0952/20-00号项目基金，项目名称为“开发用于分析公共交通车辆内部监控视觉数据的系统，以检测异常情况/行为、统计乘客数量（包括按年龄组分类）及识别乘客携带物品”；欧盟“地平线2020”计划下的AI4media项目（项目编号GA 951911），该项目为面向媒体、社会与民主的下一代人工智能研究与培训卓越中心；托斯卡纳大区INAROS研究项目（面向老年监护与支持的人工智能技术，项目编号CUP B53D21008060008）。 <strong>授权协议</strong> <em>公交暴力（Bus Violence）</em>数据集由Blees Sp. z o.o.研发并发布，采用知识共享署名（Creative Commons Attribution）许可协议，仅可用于非商业用途。