Scholarlink

<p>For the abnormal behavior risk assessment in video surveillance, this study combines the powerful feature learning capabilities of deep learning with the ability of fuzzy inference systems to handle uncertainty and fuzziness. We use a baseline network to identify pedestrian abnormal behaviors, incorporating crowd and scene features. A fuzzy inference-based risk assessment model for abnormal behavior is designed using fuzzy sets and rules, which generates fine-grained risk values and risk levels. We have constructed a new pedestrian abnormal behavior dataset and validated the effectiveness and generalization ability of the fuzzy inference system across different datasets, while also presenting visualization results. To further verify the system’s efficiency and flexibility, we conducted comparative experiments with traditional machine learning methods. The results show that the fuzzy inference system demonstrates high accuracy and inference efficiency in abnormal behavior risk level determination.</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

Risk Assessment of Abnormal Crowd Behavior Based on Fuzzy Inference Systems

  • 期刊论文

  • 发表日期：

    2025年5月19日

查看全文
  • Beihao Xi,
  • Qingkui Chen,
  • Yongcheng Zhuang,
  • Chen Zhang

摘要

For the abnormal behavior risk assessment in video surveillance, this study combines the powerful feature learning capabilities of deep learning with the ability of fuzzy inference systems to handle uncertainty and fuzziness. We use a baseline network to identify pedestrian abnormal behaviors, incorporating crowd and scene features. A fuzzy inference-based risk assessment model for abnormal behavior is designed using fuzzy sets and rules, which generates fine-grained risk values and risk levels. We have constructed a new pedestrian abnormal behavior dataset and validated the effectiveness and generalization ability of the fuzzy inference system across different datasets, while also presenting visualization results. To further verify the system’s efficiency and flexibility, we conducted comparative experiments with traditional machine learning methods. The results show that the fuzzy inference system demonstrates high accuracy and inference efficiency in abnormal behavior risk level determination.