As many as 600,000 people are reported missing in the U.S. each year, with up to 100,000 people missing at any given time. Fortunately, most people, such as lost children in large venues like shopping malls, or elderly people with dementia wandering off from hospitals or nursing homes, are located – although often after an extensive use of resources and manpower. In the past, security or law enforcement personnel would be required to manually watch video footage from multiple cameras, making their search for the missing person difficult and time consuming. However, with advances in technology, specifically AI computer vision, finding missing persons has become significantly easier.
What is AI People Tracking Technology?
Person re-identification uses full-body images of people from one camera to find and track them across other cameras. Re-identification involves matching people in different cameras based on appearance or other features such as clothing or height. For example, in a large shopping mall with multiple cameras, person re-identification technology can be used to search through the entire mall's recorded video based on a missing person’s body shape or other physical attributes from one surveillance image. The person’s movements can then be tracked in chronological order to help expediate the search.
In a typical person re-identification scenario, one camera initially detects and tracks a person, and then the system uses the information to match the same person in other cameras. However, due to differences between camera devices and camera obstructions, the individual’s characteristics could vary, and their appearance be affected by clothing, size, posture, and viewing angle.
To overcome these challenges, person re-identification systems use advanced AI algorithms and techniques, such as feature extraction and deep neural networks (DNNs), which enable the system to extract unique features from a person's appearance and match them across different cameras.
Person attribute recognition uses deep learning to extract attribute information from image data, even when a source image is not available. Attributes can include gender, age, clothing, and more.
Using the example of a lost child, when security personnel do not have a pre-existing image of the child, but know the child's attribute characteristics (gender, age, clothing, accessories, etc.), they can use person attribute recognition technology for people tracking in surveillance video. This means guardians would only need to describe the physical attributes of their missing child (such as a boy wearing a green T-shirt and blue jeans with a baseball cap), and security personnel could use person attribute recognition technology to search through all recorded video in the entire shopping mall to help locate the missing child.
In people tracking technology, facial recognition refers to the technology of identifying people through the recognition of their faces captured by cameras. When surveillance footage has captured a clear facial image, a photo of the missing child’s face can be used with facial recognition technology to greatly improve the accuracy of the search.
Due to the demographic of those who often go missing, such as the elderly, young children, and individuals with dementia, an extensive search area and time frame may be necessary to locate them. Without people tracking technology, it may be difficult to conduct an effective search and valuable, critical time may be wasted.
Using the footage of a missing person captured by a camera to identity them across all of the cameras at the location, the missing person is able to be quickly located.
Searching for a missing person in a sea of people can be difficult. Through person attribute recognition, it is possible to find individuals who match specific attributes such as gender, age, clothing, and bags. With these conditions provided by family members of a missing person, it is possible to quickly identify individuals from various camera recordings and speed up the process of finding the missing person.
A type of biometric identification technology that operates by extracting facial feature values in vector form, and then analyzing and comparing them with pre-registered facial images. With facial recognition technology, it is possible to accurately identify the facial features of a missing person from video footage and locate them more precisely.
Applications suitable for adopting people tracking technology include areas with many elderly people, young children, or high foot traffic locations.
CyberLink People Tracker perfectly integrates the above three AI technologies to provide a flexible and robust solution. For example, a major benefit of Person Re-Identification and Person Attribute Recognition is that facial recognition is not required, enabling the tracking of specific people through other physical attributes when a facial image isn’t available or if there are user privacy concerns.
People Tracker is powered by AI computer vision. Through real-time analysis of surveillance video, it can identify specific people's movements based on body shape, gender, age, clothing, hats, bags, and other physical characteristics. It can also be integrated with major mainstream video management systems (VMS).
Unlike traditional facial recognition technology, People Tracker can track people based on their physical attributes and other non-facial characteristics. Even if the facial image is blurry, specific individuals' entry and exit records can still be tracked and recorded through powerful computing power. With these capabilities, People Tracker can search for missing individuals across multiple cameras within a venue, as well as track their movements over a 48-hour period. Additionally, People Tracker can be combined with FaceMe Security, CyberLink's facial recognition solution designed for security and access control, to create a more comprehensive intelligent security environment.
The application of people tracking technology can not only be used to help find lost people, but also can be applied in other use cases, such as train stations, airports, shopping malls and other places with a large number of people. For example, in addition to assisting the application of finding lost people, it can also help police track suspects in public places, thereby improving public safety. People tracking technology has achieved good results in a wider range of fields such as security monitoring, people counting, traffic control, and more. Therefore, it’s predictable that future applications of people tracking technology will be very diverse, and is anticipated to grow significantly.