System And Method for Online Anomaly Detection In Surveillance Video

Tech ID: 20A037

Competitive Advantages

  •     Faster processing times
  •     Higher accuracy than current technologies
  •     Simple implementation into current technology

Summary

We have developed an algorithm to detect  anomalies in online surveillance video quickly and accurately.  This technology incorporates two modules, the first of which is a novel feature extraction tool that generates realistic images and videos for future frame prediction. The second module includes a statistical algorithm that uses extracted features for anomaly detections. The techniques can be implemented in connection with hardware, software or both. It also provides a decision-making strategy with the desired false alarm rate which is crucial for minimizing human involvement. 

Depiction of the System Process of Detecting Anomalies 

Desired Partnerships

  • License
  • Sponsored Research
  • Co-Development

Technology Transfer
TTOinfo@usf.edu
(813) 974-0994

Patents