AI-Based Video Security System Promises Accurate Alerts
This post is also available in: 
עברית (Hebrew)
A new autonomous AI-based video security system uses rapid neural networks and the most advanced AI technology available to provide organizations with qualified, accurate alerts about security risks in real-time to keep people safe, deter intruders and protect valuable assets.
Umbo Computer Vision’s award-winning product, Umbo Light, is a smart camera that works without physical on-premise recorders. The device stays connected vis-a-vis network connections and through the cloud. Multiple cameras can be networked and managed from multiple locations through a single Web login, which has been used more than 1.5 million times by customers. The system can be accessed from both desktops and mobile phones.
The systems are currently in use at schools, government facilities, and many commercial sites to protect critical infrastructure and safeguard the public.
The device is designed to relay data to a cloud-based, neural net based AI system in order to protect people and physical assets through interactive learning. As the device observes the environment, it is designed to provide smarter alerts over time, according to biometricupdate.com.
The development leverages proper machine learning capabilities, which results in accurate data and a lower number of false alarms, especially when dealing with complex scenes and weather conditions.
The technology is deployed by a network of 200 active system integrators and has over 500 active end-users using it across 13 physical security verticals. The system has built its intelligence by analyzing over two billion security images. According to the firm’s own research, 12 cameras or more per operator is too much visual data for a human to process. Based on this approximation, 400 million cameras would require 120 million operators to monitor them. Umbo’s algorithm allows the monitoring task to be automated and conducted in the cloud instead, accessible online and through API integration.
In addition to tracking motion, the system learns about behaviors from what it observes so it better understands a user’s needs the more it is used at a specific location. The system uses AI to sequence images and also to compare the relevance of the images to one another. The result is a temporal end-to-end solution that can alert users about critical security events or provide real-time data that can allow for even preventive action.
