Scholarlink

<p>Recognizing violent behaviors in videos is a challenging task due to the complexity of human actions and background information. In this paper, we propose STIG-Net, a spatial–temporal interactive graph framework, to address this issue. By extracting keypoints from video frames and constructing special edges based on their relationships, STIG-Net leverages graph neural networks and an attention mechanism to capture the spatial and temporal dynamics of violent behaviors. We evaluate our method on five benchmark datasets, achieving outstanding accuracy and AUC scores. Our approach effectively reduces the influence of background information and demonstrates high recognition rates, as evidenced by ablation studies. The source code and datasets are openly available at <a href="https://github.com/Estancys/STIG-Net">https://github.com/Estancys/STIG-Net</a>.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

STIG-Net: a spatial–temporal interactive graph framework for recognizing violent behaviors in videos

  • 期刊论文

  • 发表日期：

    2025年2月7日

查看全文
  • Xinbiao Lu,
  • Yisen Chen,
  • Yudan Chen,
  • Xing Gao,
  • Tieliu Yang,
  • Guiyun Chen

摘要

Recognizing violent behaviors in videos is a challenging task due to the complexity of human actions and background information. In this paper, we propose STIG-Net, a spatial–temporal interactive graph framework, to address this issue. By extracting keypoints from video frames and constructing special edges based on their relationships, STIG-Net leverages graph neural networks and an attention mechanism to capture the spatial and temporal dynamics of violent behaviors. We evaluate our method on five benchmark datasets, achieving outstanding accuracy and AUC scores. Our approach effectively reduces the influence of background information and demonstrates high recognition rates, as evidenced by ablation studies. The source code and datasets are openly available at https://github.com/Estancys/STIG-Net.