ZONE POSITIONING WITH THE FACE ID SYSTEM
The principle of people positioning in any premise based on video cameras is simple by its structure. The cameras should be installed in such a way that the faces of the passing stream of people outline at a distance of up to 8 meters. The best place to install the camera is the aisles. This way the number of cameras will be minimized. People will be well-localized in the camera’s coverage area. First we need to understand what the Face ID system is. Primarily, this is a trained neural network that is created specially for such a task. Secondly, it is a face-recognition server that processes video streams from cameras.
The server can be integrated into the camera but such equipment is rather expensive because image handling requires a powerful processor. A neural network distributes feature vectors and finds geometric patterns of the face. This process is called neural network training. A trained network hereinafter can determine a person with a probability of up to 98%.
How Does a Neural Network Identify a Face?
- The recognition system chooses a separate frame and analyzes whether there is a person’s face on it.
- The neural network selects from 5 to 1000 control points on the frame with the image of the face. These points determine how the face is located in the space. The Merusoft Face ID engine uses 82 points.
- Then, the Face ID engine converts the image into a cipher, describing a large number of parameters specified in the neural network. These parameters are formed by the neural network right after training. Because during training, two images were loaded at the input of the neural network and then only the correct one was indicated.
- The server compares the cipher received in the image processing with the reference version (specified patterns). If a match is found, the system gives the result of the identified person.
The use of cameras has gained mass adoption. The cost of the Face ID software is reduced every year due to high competition in this segment.
How Face ID works?
Features of RTLS Solutions with Face ID Technology:
We are developing RTLS applications using Face ID technology on the Merusoft Tracking platform. To develop IPS (Indoor Positioning System), we use any installed cameras, as well as Face ID engines. The software can be installed on the client’s infrastructure without connecting to the Internet that is important for industrial enterprises. Each user receives a unique identifier in the Merusoft Tracking system. It is used for integration with the Face ID system. Cameras are attached to the coordinates of the floor plan or map. Thus, the location of people on the map is foregrounded according to the data obtained from video surveillance cameras.
Areas of the positioning system using CCTV cameras:
- Chemical industry
- Food industry
- Exhibitions
- Robotic production with conveyor lines
- Warehouse complexes
- Mobile systems for monitoring the movement of fire brigades
- Offices and business centers
Using cameras, you can quickly deploy a positioning system to conduct research and analyze the movement of streams of people.
The presence of cameras and the Face ID engine is necessary for the positioning system to work on the Meruosft Tracking platform.
