The development of artificial intelligence has been fascinating over the past decade, but the best has not yet come. As we prepare for entry into 2018, there are many exciting developments taking place, especially video surveillance of enhanced artificial intelligence.
For years, researchers have clearly demonstrated advances in machine learning and artificial intelligence. Whether it’s IBM’s Watson winning a game against the world’s brightest, a Chinese platform outquiring humans on IQ tests or Google’s AI wrote dozens of his poems The example demonstrates the greatness of artificial intelligence techniques.
However, to date, artificial intelligence has done more than help. However, the times are changing. In addition to major breakthroughs in medicine and health care, the security industry also saw some positive developments. Specifically, new developments will improve video surveillance.
Artificial intelligence overcomes human shortcomings
Entrepreneur Bob Bechile explains: “One of the problems with monitoring operators is the fatigue of attention.” The brain naturally alternates between distracted and distracted. Distractions can have terrible consequences in surveillance. So, if we have an undisturbed surveillance system, a surveillance system that works with humans to reduce errors? This is the AI’s commitment to video surveillance.
While video surveillance technology has historically been documented and allows people to observe it when they want it, it has never told a human operator, “Hey, that does not look right!” “In addition to motion detection, often False alarms are generated and none of the features can actively make up for the psychological defects of the human brain. ”
Just last year, machine vision development company Movidius and security hardware maker Hikvision announced their intent to cooperate and produced a new camera equipped with deep neural network artificial intelligence, which will notify you when something is happening , And provide real-time alerts.
A real example is monitoring a store and noting that the same person strolls and / or returns to the same place several times over the same place within a given time frame, possibly to steal property. The camera will find such behavior – something a human camera operator may not have noticed – and warns of further investigation.
Superior facial recognition
To find out who the murderer is, shots are often hard reviewed. Unfortunately, a person can only confirm his identity if he or she knows the person on the screen. Advanced artificial intelligence software like MegaFace hopes to change this situation.
The MegaFace project is a project at Washington University. It was trained on a dataset with nearly 5 million images and about 672,000 people from Flickr. In the test, the accuracy of this system was 75%, matching two pictures of the same person on different datasets of one million facial images. If you have a chance to guess 10 times, the accuracy is increased to 90%.
Obviously, facial recognition systems have many potential values that can be combined with video surveillance. It can help law enforcement officers find individuals in the crowd, which may help to track offenders and even prevent the crime from happening.
Real-time response to crime
Hope that these updated artificial intelligence monitoring system can be found in real-time crime. For example, the Movidius system can actually discover unattended objects. Practical application will be the airport and train station. The cameras can identify unattended bags, track the time left by them and warn of on-site security personnel to observe suspicious activity.
Another example is a camera system that is trained to identify body language and movements used by the thief before theft. When these behaviors are identified, the security team can deploy personnel to check the situation.
AI and future video surveillance
Artificial intelligence has great potential in many industries and fields, but it is hard to ignore its shift in security. Whether it’s business monitoring stores, homeowners protecting their personal belongings, or law enforcement officers identifying criminal activity, the applications are numerous.
It will be very exciting to see developments in the coming years.