Crowd Behavior Analysis
Through its world-leading AI technologies, NEC is continually pursuing the evolution of its advanced and diverse physical security solutions. For public spaces, such as cities, airports, and stadiums, NEC prevents the occurrence of accidents and incidents through early detection of dangerous behavior and suspicious vehicles from video surveillance images. One of NEC’s solutions is Crowd Behavior Analysis, a technology that detects congestion and unusual phenomena from surveillance videos in real time, while maintaining personal privacy. Crowd Behavior Analysis Technology can analyze crowd footage and estimate the number of people, it can be used to prevent crime, and also to respond quickly and guide residents to safety in an emergency or disaster.
When a large-scale facility or public space attracts a large amount of people, there is always the risk of an unexpected incident taking place, including accidents and acts of crime. As a result, a wide range of heavily trafficked facilities are equipped with security cameras in order to help prevent accidents and to minimize possible damages. However, in a crowded environment, where hundreds of people are interacting in a complex mixture of activity, it is difficult for conventional technologies to quickly detect and identify the characteristics of a disturbance.
NEC Crowd Behavior Analysis Technology analyzes the influence that an unexpected incident, or the signs of an incident, has on the behavior of a crowd in order to accurately identify changes in the crowd's behavior. As a result, existing security cameras can be used to detect disturbances, such as a sudden change in the flow of foot traffic or a crowd formed around a fallen individual, and to accurately estimate the degree of congestion, even in an extremely crowded environment, such as an event venue or train station during rush-hour.
Accurate understanding of crowd conditions
This technology creates pseudo-images through simulations that demonstrate a wide range of different conditions, such as varying degrees of crowd congestion and each individual's behavior. Image recognition technologies then employ a proprietary algorithm to match and analyze security camera images against pseudo images. This enables crowd conditions to be understood with a high degree of accuracy even when people in the images appear to be heavily overlapped.
Fast and accurate analysis of changing crowd conditions
Special focus is placed on the behavior of people in the vicinity of any unusual occurrence. If, for example, two individuals begin to quarrel in a public space, the technology closely analyzes the behavioral patterns of people surrounding the incident, such as temporary pauses or large gatherings. Moreover, the level of an incident can be estimated based on a pre-determined threshold value that helps measure the magnitude of a crowd's change.
Focus on the movement of surrounding people
When someone falls in a crowded place, it is very difficult to pinpoint just that person in a surveillance video, and it delays the initial response of security personnel as well. Crowd Behavior Analysis can analyze such group changes in real time and generate an alert. It was this shift of thinking to focus not on identifying an individual for which to generate a warning but on the people around them that produced this technology.
Efficient and reliable security
Crowd Behavior Analysis detects abnormalities and creates alerts, so that risks are not overlooked and trouble is prevented before it occurs. This also leads to optimal deployment of security personnel and increases the efficiency of security.
Conventional technology analyzes the faces or individuals. When bodies overlap, it is difficult to get an accurate analysis. On top of that, there was the risk of infringing upon people's privacy. With Crowd Behavior Analysis, people are analyzed as groups. This enables accurate analysis even in crowded situations where people overlap in the camera. Moreover, individuals are not identified, so privacy is ensured.
Predict urban dangers
Crowd Behavior Analysis technology holds further potential to ring in the future. It will watch out for the safety and security of the city in an integrated fashion by combining and aggregating information from acoustic, vibration and other sensors in the town. We will connect the world's desires with the future.