The VCA Approach
Our approach to AI and Deep-Learning (DL) is very different to many other companies in security analytics. As an early pioneer of video analytics (since 2007), we understand the real issues on analytic reliability and what causes false alarms.
We also understand what our clients want and put simply that is “a significant reduction in false alarms and ease of set up”.
We have therefore concentrated the development plan on satisfying our customers demands by making false alarm reduction and ease of set up the key objective.
How do we do this?
Our first Deep-Learning (DL) module release is focused on “Event Filtering” and is ideal for sterile zones and perimeter protection. This is the DL Filter module
The deep-learning based filter can be used to augment the current VCAcore classification engine based on calibrated scene or be used as a stand alone module.
It more accurately distinguishes between true events and false positives (e.g. spiders, car headlights, environmental changes) It does not require camera calibration, which decreases the installation and maintenance burden.
We will release our second DL module early in 2019. This module is ideal for busy scenes and it utilises our high accuracy tracking engine for accurately tracking humans in busy, unconstrained dynamic scenes.
It is ideal for retail Intelligence, security at high density scenes such as airports, travel hubs, and Casino’s.