OneBerry

ARVAS

ARVAS is an AI-assisted video anomaly detection system that requires no input of rules or pre-configuration. It continuously analyses video streams and highlights any abnormal behaviour to operators in real time. With self-learning capability, ARVAS can adapt to different environments and detect unlimited range of anomalies across all sectors. ARVAS is the future of smart surveillance. 

Jump to

Availability

  • Americas
  • Asia Pacific
  • Europe
  • Middle East And Africa

OneBerry Arvas Solution Overview

ARVAS (Abnormality Recognition Video Analytic System) is Oneberry's (formerly VI Dimensions) patent pending Abnormality Detection Technology. ARVAS analyses video data to uncover patterns and find abnormalities that would otherwise have gone unnoticed.

Key market application

ARVAS uses unsupervised machine learning which does not require users to specify the rules for finding abnormalities, reducing the manpower required and increasing efficiency. All that is required is a few hours worth of videos to get the system started and deployed. Learning will continue in real-time (along with detection) as the system continually observes the scene. Alarms can then be bookmarked and sent to Genetec Security Centre for further processing or human validation.

Traffic Monitoring

ARVAS can be applied in monitoring traffic on a large scale. No rules are required to be set for each camera to define normal traffic patterns. Such patterns can be automatically picked up by the software. 

Traffic violations related to motion patterns can be alerted especially for cases such as wrong-way driving, accidents, going on road shoulders, etc.

Monitoring Infrastructure

It is impossible for anyone or a group to monitor more than a few hundred cameras in real-time in crowded spaces, especially in busy places such as stadiums, traffic stations, airports, etc.

ARVAS has shown success, for this particular context, in detecting certain cases involving infrastructure monitoring. For example, cases where people are climbing over gates for fare evasion, climbing over structures, running, panicking crowd, etc.

Features and benefits

  • Unsupervised Machine Learning - Uses the latest and most advanced machine learning techniques
  • Autonomous Real-Time Detection - No need to set up rules
  • Unlimited Variation of Behaviours - Behaviours are no longer limited by fixed rules
  • High Scalable - Scalable for hundreds or thousands of cameras
  • Not scene dependent - Can be applied to a wide variety of camera scenes
  • Event Clustering - The system can cluster similar events together

Solution architecture