Advanced iOmniscient Technologies

Advanced Techonology * IQ-Smart City

Just like humans, different Video Analysis systems have varying levels of intelligence.

In view of the considerable confusion associated with defining intelligence for video-based surveillance systems, a rating system was developed for various types of intelligence capabilities. The criteria used to define the level of intelligence, essentially relates to how much “thinking” or analysis the system is capable of and to the level of difficulty and complexity of its thinking. The system is similar to that used for measuring human intelligence. Just for reference, the average iQ of the human population is 100. Half of the population has an iQ of more than 100 while the other half has less.

A surveillance system with a rating below 90 would be classified as a system with very basic or limited intelligence. Between 90 and 110 the system has average intelligence. At an iQ of 120 the system can be considered to be very smart and above that, the system would be in the genius category.

The chart shows the iQ level of various functions. There is a direct correlation between the complexity of analysis provided by systems and the number of companies in the market place that can provide the capability. There are now hundreds of suppliers that provide systems with limited intelligence. At the average level of intelligence there are about a dozen companies worldwide that can provide a robust system.

There are only two or three companies worldwide with offerings that meet the criteria for “Smart” and “Very Smart” systems. At the “Genius” level, because of its international patents, iOmniscient stands alone in providing products in this class.

When applied to video based surveillance systems, storage and retrieval systems would be positioned at the bottom of the scale as having very limited intelligence. At the next level of intelligence, we have image enhancement systems which are systems that make the image clearer to see. An auto focus function or a system that eliminates camera shake would fall into this category. There is some basic intelligence involved here but it is still very simple.


iQ-Rating Chart

Advanced Techonology - IQ Rating Chart * IQ-Smart City

The rating system above is now widely used by some consultants and security analysts as a guideline for measuring the intelligence of video based systems. Systems with the capacity of simple motion detection using image differences would be classed as being of limited or basic intelligence.

At a slightly higher level of intelligence, the system should be smart enough to differentiate between objects using simple techniques and then to recognize simple behaviors, such as someone crossing an imaginary line. At the next level, there are systems that can analyze more complex behaviors such as running or loitering. This level of technology is available from several major suppliers and would be classified as having “Average” intelligence.

A smarter system would be able to count people and vehicles accurately. In fact, the same system should be able to classify objects (e.g. trucks, cars and bikes) and count each category individually.

An even more sophisticated system is capable of more complex behavior analysis, such as the ability to detect slips and falls.

This would be considered as a smart system. People can fall down for many reasons. They could have been assaulted; they may have had an accident; they may just slip and fall. The system detects such an event and enables resources to be deployed to help the person. Alarms will not be activated if a person bends down to tie a shoe lace but if the person falls down suddenly, the system will raise an alarm.

This application has been found to be useful in reducing insurance claims. People who get immediate attention and care are less likely to initiate litigation than those who are ignored. More importantly, it ensures that if there is an accident, the victims will receive immediate help. Genius systems go beyond the abilities of motion-based systems. They can detect stationary objects in a crowded scene with very long detection times. They can also detect objects that might be invisible to the human eye. iOmniscient's advanced and patented technology enables the detection of abandoned objects in crowded spaces and complex environments including outdoor environments. It is the only software vendor that has this proven capability.

Cities have a security challenge in detecting abandoned objects in crowded public areas. Given the high volumes of people passing through these areas, there is a very high probability that someone will put a bag down at some point. But these bags are probably not dangerous. They are just the normal activity of people moving about in a public space.

The CCTV surveillance system should be able to ignore all the bags that may have been put down for 2 minutes or even 5 minutes and catch only those which have been abandoned for a longer period of time - say 10 minutes or longer. Even 10 minutes may be a too short time period but one has to strike the right balance between detecting suspicious objects and ignoring the thousands of bags that everyone puts down momentarily when they are in a public area.

During this time ten, maybe up to hundreds of people may walk in front of the abandoned bag or object. They may stand in front of it and may totally obscure it from time to time. If the bag or object is still there after the defined detection period, the system needs to be able to detect it and raise an alarm. Theft of cars or items from cars and theft from stores can be a major problem, too. The same technology can detect when stores or vehicles are broken into and raise an appropriate alarm. Vandalism and graffiti can be a problem at bus stops and at public buildings. The technology used to detect abandoned bags can also detect graffiti and vandalism even if it occurs in a busy environment with people and vehicles moving about and partially or totally obscuring the problem area for a significant proportion of the time.

It should also be noted that some companies can offer point solutions, e.g., the detection of one type of behavior. To differentiate between these systems and those that provide a comprehensive solution, the category of iQ-Infinity was defined. In this class, the system has the ability to perform all the analyses described above at the same time on the same camera.


Immediate Response with “Jump to Event”

Advanced Techonology - Immediate Response with “Jump to Event” * IQ-Smart City

The old method of using Pre and Post Alarm video to determine what happened during an incident has become obsolete because in today's world, the incidents themselves are becoming much more complex.

The required information can now be available in real-time from iOmniscient. The “Jump to Event” capability takes information retrieval to a new level of sophistication.

In simpler systems, a few seconds of video may be retrieved showing the moments just before and just after the event.

This works when the event happens at a fixed point in time, such as a person jumping over a fence. The user can note the moment that this event occurred and store footage from just before and just after this event. This is referred to as Jump to Alarm and virtually all iOmniscient systems can provide this capability.

If the event takes place over an extended period of time, the requirements for Jump to Event can become very complex. For instance, if a bag is abandoned in a crowded scene with a detection time of 10 minutes, the bag has been identified as abandoned and an alarm is triggered by the system after 10 minutes. If the user looks at the video for 30 seconds before and after the alarm, the bag would be seen but without a history of how it got there.

iOmniscient has developed a unique Jump to Event capability associated with an iQ-140 level system that goes way beyond this. It enables the user to jump back to a pre-defined time before the event started. Hence, in this example of the abandoned bag, the user can jump back in time not just before the bag was detected but before it was first brought into the scene and abandoned with the press of a button. The user can immediately review both the event and the preceding period of video footage and gain important information about who brought the bag in and the person's current location if he is still within camera view.

Now consider a scene where there might be ten suspicious bags in the area being viewed. The Jump to Event function should enable the user to click on any one of these bags and “jump back” to the moment when that particular bag was brought into the scene and abandoned. The user then has the option of archiving the event for later review or discarding the alarm information.

None of the simpler data storage and retrieval systems has such a capability as it is very dependent on the intelligence required to find the bag in the first place. Only a system that can operate at the iQ-140 level and above can actually have a sophisticated Jump to Event function similar to the one described here. This is the domain in which iOmniscient has international patents.


Recognition Technologies

All the technologies described so far are detection technologies. They analyze a wide scene and detect behaviors of humans or vehicles in the scene. There is a parallel Video Analytic technology for the recognition of humans and vehicles.

Face Recognition in a Crowd

The most powerful new technology introduced by iOmniscient within the last 12 months is the world's first Face Recognition in a crowd. Using relatively low resolution video (the system requires only 22 pixels between the eyes) for recognition, the system can recognize people who are non-cooperative, match them with data bases and provide a very useful tool for catching persons of interest in general.

Advanced Techonology - Recognition Technologies * IQ-Smart City

The most powerful new technology introduced by iOmniscient within the last 12 months is the world's first Face Recognition in a crowd. Using relatively low resolution video (the system requires only 22 pixels between the eyes) for recognition, the system can recognize people who are non-cooperative, match them with data bases and provide a very useful tool for catching persons of interest in general.

iOmniscient's Face Recognition System can operate in an uncontrolled environment requiring only 22 pixels between the eyes to perform recognition. The system can cope with variations in pose and expression, head gear, spectacles and facial hair. The system is not designed as a biometric access control system but is rather primarily used in surveillance. With its ability to match faces with multiple large databases, the system provides a far higher accuracy than the alternative in such situations which is a manual system.

In addition to matching faces with multiple lists of people, it can be used to track a person. If person A has been seen on a particular camera and tagged, the system can determine the time and on which other cameras A has appeared. This is perhaps the only automated technique available today for tracking people on non-overlapping cameras. In addition to detecting and tracking potential criminals, it could also be used to find lost children at a crowded event. The system can be used for customer service to see how long it takes a particular person to travel from Point A to Point B.

It can be used for recognizing witnesses at an accident and to determine where they went if they had not remained at the scene when the police arrived. The demographic information that can be extracted from the Face Recognition system can have multiple purposes to ensure the safety and security of citizens.

Multi-lingual License Plate Recognition

iOmniscient's License Plate Recognition system can recognize plates of over 100 countries. It does not require any external trigger, e.g., motion induction loops, to recognize that a vehicle is in view and needs to be recognized. It is all done by video. Operating at high accuracy even at high speeds, such an integrated License Plate Recognition system is critical in any Smart City implementation.

Driver Match

The Face Recognition and License Plate Recognition system can be used together. In public car parks, vehicles may sometimes be driven out by someone other than the driver who brought the vehicle in. Very often, this is a symptom of a vehicle theft. Experience shows that it is often insufficient to use humans at checkpoints to detect such thefts of vehicles. The Driver Match system recognizes the number plate of the vehicle being driven as well as the face of the driver and checks if the driver is authorized to drive that vehicle.