Scholarlink

In today’s world to identifying crowd unusual activities stands as an essential requirement to protect both smart cities and operate successful events together with operating transportation systems. When utilized for big datasets and variable crowd patterns traditional recognition approaches show difficulty in operation. The deep learning system uses Convolutional Neural Networks and Long Short-Term Memory networks to detect problems within heavily populated settings. The methodology processed ShanghaiTech along with UCF-Crime datasets and achieved 94.2% accuracy that surpassed Autoencoders and GANs models. Real-world surveillance systems use scalable solutions developed from theoretical algorithms which link theoretical discoveries to practical engineering applications.

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

Deep Learning for Crowd Anomaly Detection: A Hybrid Spatiotemporal Approach for Public Safety

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Dharmesh Tank,
  • Sanjay G. Patel

摘要

In today’s world to identifying crowd unusual activities stands as an essential requirement to protect both smart cities and operate successful events together with operating transportation systems. When utilized for big datasets and variable crowd patterns traditional recognition approaches show difficulty in operation. The deep learning system uses Convolutional Neural Networks and Long Short-Term Memory networks to detect problems within heavily populated settings. The methodology processed ShanghaiTech along with UCF-Crime datasets and achieved 94.2% accuracy that surpassed Autoencoders and GANs models. Real-world surveillance systems use scalable solutions developed from theoretical algorithms which link theoretical discoveries to practical engineering applications.