Scholarlink

In the rapidly evolving landscape of video anomaly detection, we present an innovative and robust approach that leverages the fusion of VGG16 with a denoising autoencoder. Our model entails extraordinary performance, obtaining a remarkable precision of 91% through detailed data preparation, model training, and thorough assessment. It demonstrates outstanding flexibility and accuracy in recognizing abnormalities within video streams by successfully tackling the complexity of real-world data, including noise and false labeling. This work advances the field of anomaly detection while simultaneously demonstrating the potential of deep learning approaches to enhance security and surveillance in a data-driven society. With an overall accuracy of around 80%, recall of 80%, precision of 91%, and F1 score of 85%, the performance of the hybrid model across many classification measures are balanced.

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

Enhancing Video Anomaly Detection: A Hybrid Model Leveraging VGG16 and Denoising Autoencoders for Superior Precision and Robustness

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • M. Lakshmi Prasudha,
  • Vidyullatha Sukhavasi,
  • Kandula Neha,
  • Poonam Shaylesh Lunawat

摘要

In the rapidly evolving landscape of video anomaly detection, we present an innovative and robust approach that leverages the fusion of VGG16 with a denoising autoencoder. Our model entails extraordinary performance, obtaining a remarkable precision of 91% through detailed data preparation, model training, and thorough assessment. It demonstrates outstanding flexibility and accuracy in recognizing abnormalities within video streams by successfully tackling the complexity of real-world data, including noise and false labeling. This work advances the field of anomaly detection while simultaneously demonstrating the potential of deep learning approaches to enhance security and surveillance in a data-driven society. With an overall accuracy of around 80%, recall of 80%, precision of 91%, and F1 score of 85%, the performance of the hybrid model across many classification measures are balanced.