Robust Real-Time Violence Detection in Video Using CNN And LSTM

  • Unique Paper ID: 163048
  • Volume: 0
  • Issue: no
  • PageNo: 256-260
  • Abstract:
  • A major part of law enforcement and public safety is the detection of violent events in surveillance systems. The speed, accuracy, and adaptability of violence event detectors across a range of video sources in various forms provide measures for measuring their success. A number of studies focused on speed, accuracy, or both when detecting violence, but they ignored to account for adaptability across various types of video sources. In this paper, we suggested a deep-learning based real-time violence detector. CNN serves as a geometric feature extractor in the proposed framework, while LSTM is used as a time-based relation learning technique with a focus on the three factors (accuracy, speed, and overall flexibility). The advised model reached 98%.

Cite This Article

  • ISSN: 2349-6002
  • Volume: 0
  • Issue: no
  • PageNo: 256-260

Robust Real-Time Violence Detection in Video Using CNN And LSTM

Related Articles