• 29 Tai Seng Avenue, #05-01, Natural Cool Lifestyle Hub, Singapore 534119
  • (+65) 6818 1818
  • (+65) 6822 1327

Abnormality Recognition Video Analytic System (ARVAS)

Abnormality Recognition Video Analytic System (ARVAS)

Abnormality Recognition Video Analytic System (ARVAS) is a flagship product of Vi Dimensions. It differentiates from other conventional video analytics software in that it employs a selflearning algorithm which does not require the surveillance operator to pre-define any rules for real-time detection. The system uses GPU for massive hardware scalability, allowing Artificial Intelligence and Machine Learning to be effectively deployed for auto-detecting abnormal events in real-time. The use of Deep Neural Networks also allows the human operator to provide feedback to the system in order to manage and prioritize the alarms.

How ARVAS works

Instead of defining rules, we look for deviant activity patterns. The system can now discover all kinds of abnormal behaviours (even those we do not
know beforehand).

How ARVAS Addresses Issues
Detect Abnormal Events in Real-time

Surveillance videos no longer just for post-event forensics

Unsupervised Machine Learning

No need to predefine any rules for detection

Detection of Unknown Behaviours

Adaptation to constantly changing scenes can detect unknown behaviours

Improved Automation

Increases efficiency of 1 operator to monitor and manage 1 thousand cameras

Highly Scalable

Scalable for thousands of cameras

Deep Neural Net Event Clustering

Enables user feedback for alarm prioritisation and management