Traditionally AI based computer vision systems were only designed for empty spaces as crowded scenes generate many challenges. Objects and people in a crowd may be quite far away and may only be visible at a very low resolution and they may be partially or, at times, totally obscured making them very difficult to be recognized by systems trained using traditional methods. Further they are usually only designed to detect one behaviour at a time.

iOmniscient systems are designed to deal with such complex situations. Its internationally patented capabilities can handle a variety of such complex situations that can arise in crowded scenes.

The most serious safety issues arise when there are subtle changes in the behaviour of the crowd.  Understanding these sudden changes in the direction of the flow of the crowd can help authorities predict potential consequences. The sudden formation or dispersal of a crowd can also indicate other underlying problems that may not have otherwise been detected.

Detecting the underlying causes of a change in crowd behaviour can also help authorities manage the response. Detecting screams, gunshots, explosions and gas leaks can help pre-empt. The way a crowd might respond to a theft or act of vandalism is different to how it may respond to a fire.

The system may detect one or all of these symptoms and must be able to:

• Understand what is happening
• autonomously determine the type of response that is required,
• find the nearest appropriate responder and
• make them aware of the situation telling them where to go and what to do

Understanding the combination of different behaviours and their sequence and the ability to generate an autonomous response is what makes iOmniscient different.

Long Experience with Crowds

The iOmniscient system was specifically designed with crowds in mind and has been deployed in projects across 70 countries in the densest of crowds for three Olympics, one FIFA tournament, one Commonwealth Games in theme parks and pilgrimage sites which are always crowded.

The algorithms developed are unique and computing light so they do not require GPUs. This significantly reduces their cost but also their power utilization. Typically, an iOmniscient system will have a 16x lower carbon footprint making it much more sustainable.

iOmniscient systems are also designed to use legacy low resolution cameras so the infrastructure does not need to be upgraded.

Savings in Infrastructure pays for the system

iOmniscient systems are designed to reduce infrastructure costs. If 2 mp cameras are used the resultant savings are 90% for both storage and network bandwidth. No GPUs are every required as iOmniscient’s AI algorithms have evolved beyond traditional Deep Learning.  The design allows operational and staffing costs to be greatly reduced. The value proposition from iOmniscient is that it can implement intelligent systems at a lower cost than a recording system.

This ability to cope with crowds, provide an autonomous response and reduce infrastructure costs for AI systems is protected by a large number of international patents and is therefore not available from anyone else.