Types of Video Analytics

According to Thomas L. Norman in Effective Physical Security, there are three common types of video analytics:

• Fixed algorithm analytics
• Artificial Intelligence (AI) learning algorithms
• Facial recognition (FR) systems

Both fixed algorithm analytics and AI learning algorithms determine unwanted or suspicious behavior in the field of view of an IP camera, and push that information as a notification to the administrator’s computer.

Fixed Algorithm Analytics

The only difference is that fixed algorithm analytics look for specific instances, called “events” such as:

• Crossing a line
• Counting people
• Identifying a license plate
• Detecting motion
• Detecting objects left in a specific area
• Detecting long queues or lines (ex. in supermarket payment lanes)

Artificial Intelligence (AI) Learning Algorithms

Artificial intelligence (AI) learning algorithms collect information, and with the help of the administrator, learns which events to look out for.

It is similar to a child learning from a parent. The child “absorbs” all information like a sponge, and the parent identifies and corrects wrong behavior. This is how the system learns and becomes more sophisticated, reaching a point wherein it identifies events correctly, with complete details.

Although AI learning algorithms take time to be useful, they are the most flexible and actionable when it reaches sophistication.

Read More: Smart Command Center

Facial Recognition (FR) Systems

Facial recognition (FR) systems use IP cameras to collect a specific type of data based on people’s faces. It creates a database of people’s face data, and thereafter, compares detected faces seen by the IP camera with it.

It can be categorized as a biometric, like fingerprints, irises, and voices for identification of individuals.

Facial recognition is commonly used for access control, but has many other applications, such as:

• targeted advertising
• making customized experiences for customers
• people tracking in surveillance systems for case investigation
• contactless time & attendance
• secure payment / transactions

References:

1. Effective Physical Security: Video Analytics by Thomas L. Norman CPP, PSP, CSC
2. Face First: Facial Recognition Use Cases