system profiler
With more and more channels of social security video surveillance, more and more complex systems and more and more data, how to effectively collect useful video data information becomes more and more urgent. The traditional video surveillance is to query the video afterwards, and the video AI behavior intelligent analysis and early warning system realizes the synchronous early warning function, which makes the monitoring intelligent.
The solution of video AI behavior intelligent analysis and early warning system is a set of monitoring and early warning system that can run independently. Just add an intelligent early warning analysis host on the basis of the original monitoring system, and the intelligent early warning function of the monitoring system can be realized. When preset events occur in the monitoring field of vision, such as people falling to the ground, asking for help, invading, fighting, not wearing helmets, mobile phones, scissors, knives, etc. , they will immediately trigger the alarm and enter the emergency plan mode.
Technical introduction
Video AI behavior intelligent analysis and early warning system is a set of systems integrating behavior recognition technology, face recognition technology and object recognition technology. Among them, human behavior analysis technology is a visual analysis algorithm based on AI neural network, which outlines the human structure according to the pictures taken by the camera, analyzes the human motion trajectory and identifies various abnormal behaviors and actions of people. Article identification and early warning uses artificial intelligence technology based on user's needs to test and learn the articles in the video repeatedly, so that the system can identify and give an alarm when the articles appear in the system setting area.
Three technologies:
1. Behavior identification: identify abnormal behaviors of people, and make early warning and marking.
2. Wear identification: identify objects worn on the head, objects worn on the back of the body, and objects worn on the hands.
3. Object recognition: Identify the monitored items related to social security, such as cars, motorcycles, bicycles, scissors, knives and furniture.
Human behavior analysis project
Video AI behavior intelligent analysis and early warning system is based on the detection of human body structure in video images, which adds four dimensions of time, action, number of people and direction, and expands more detection events. According to the use environment, design and deploy analysis rules for various scenarios, such as: illegal entry, falling to the ground, asking for help, fighting, gathering people, leaving the post, sleeping on duty, getting out of bed, getting up, going to the toilet overtime, following the toilet, being alone, etc.
Project identification analysis project
Items that need to be detected and identified include whether to wear a helmet, use a mobile phone, knives, motorcycles, bicycles, cars, boxes, whether to smoke, fireworks, work clothes, and misplacing.
Analysis project function introduction:
1. Breaking-in event: The definition of breaking-in behavior can be flexible, and it can be applied to all scenes where it is forbidden to break in at will or in a specific time period. Intrusion behavior can be very practical through effective zone setting, invalid zone setting and time period setting. Mostly used around the fence, entrances and exits, key workshops, restricted areas, etc.
2. Fall/fall: defined as falling when the head, hips and feet are on the same plane and parallel to the ground. You can set the designated area or all-terrain area, set the alarm time period, and choose high sensitivity or low sensitivity. It can be used in public areas such as nursing homes, hospitals, entrances and exits.
3.SOS event: When a person encounters an emergency, or his hands are raised on his head, waving or reaching over his head repeatedly, he can send out an SOS signal. It can be applied to campus, nursing homes, supervision places, etc.
4. Off-duty detection: every fixed duty post must be attended. When the duty post detects that there is no one in the area beyond the set time, it is defined as off-duty behavior. Such as duty room, entrance and exit posts, etc.
5. Sleeping on duty detection: defined as sleeping when it is detected that people lie on the table for a long time or close their eyes for a long time. Such as offices, monitoring centers, duty rooms, doorways, etc. For more details about behavior analysis, please consult Iridium Company of Wei Yun.
With the development of national smart security and smart city construction, intelligent monitoring and early warning response, as an essential product, will be more widely and fully applied.
This paper is edited and published by infrared Wei Yun ultra-wideband indoor positioning system.