The Face recognition analytics. The camera placement requirements
The Face recognition analytics is a technology effectively used across various spheres. It has a wide range of applications, for example, identifying a visitor and reacting to this type of event, controlling access to a territory, improving security services operations, and much more.
For this type of analytics to be performed, an IP video camera with relevant specifications should be properly placed and set up, taking into account the location and operating conditions, in order to receive images of the required quality. A database of images for matching is also required.
In this guide, you find the requirements for the main specifications of the IP videocamera, the placement of the camera, and the images in the database to ensure the performance of the Face recognition analytics.
Videocamera specifications
When selecting an IP videocamera, you should take into consideration the requirements for images to be received, the camera specifications, the installation requirements, the control zone peculiarities, and the location and operating conditions.
For the Face recognition analytics to be performed, an IP videocamera should have the following specifications:
Specifications | Measurement |
---|---|
Matrix It is a key component of a camera that defines image quality. The value of this component reflects the size of the matrix diagonal in inches. | 1/3 inches and larger, with a high degree of sensitivity to light. |
Lens aperture ratio (light sensitivity) Plays a crucial role in overall image quality. Lens aperture – is the maximum light capacity of a lens, which determines how much light can pass through to the camera sensor. The F-number means the maximum size of the aperture. The smaller the F-number, the more light the lens can gather. | F1.4 and better. |
Lens aperture With this camera component, you can set the amount of light that goes through the lens and control the focus depth. Located within the lens itself, the aperture regulates light intake by altering the size of the opening. | With adjustable settings (it’s important to be able to set the limits of the lens aperture opening and closing). |
The requirements for face images in the database
For the Face recognition analytics to be performed, an image should be stored in the database to match the image received from the camera. The ISO/IEC 19794-5:2013 standard specifies detailed requirements for images.
A short list of the facial image requirements is as follows:
![image 16.png](../../../__attachments/5018158023/image%2016.png?inst-v=5e52fa89-af48-462a-9bad-31c36e430795)
The database image requirements
Resolution: 320х240 pixels.
Uniform face illumination.
Even, neutral background.
Face placement on an image is according to the set parameters.
The interpupillary distance is not less than 60 pixels.
The camera placement requirements
![image 16 (1).png](../../../__attachments/5018158023/image%2016%20(1).png?inst-v=5e52fa89-af48-462a-9bad-31c36e430795)
Camera placement
Distance
An optimal distance from a camera to the recognition area is no more than six meters.
Central observation axis deviations
The central observation axis should not diverge more than 15° from the facial plane in both horizontal and vertical planes. The deviation conditions should be met at each recognition point.
![Угол установки.png](../../../__attachments/5018158023/%D0%A3%D0%B3%D0%BE%D0%BB%20%D1%83%D1%81%D1%82%D0%B0%D0%BD%D0%BE%D0%B2%D0%BA%D0%B8.png?inst-v=5e52fa89-af48-462a-9bad-31c36e430795)
A video camera’s vision angles
The depth of field
The camera depth of field should cover an entire control zone. The greater the depth of field, the larger part of the frame will be clear. The depth of field depends on the focal lens and the value of the lens aperture ratio. We recommend calculating the camera depth of field in advance (which is particularly relevant for long-focus lenses) and getting a camera with adjustable lens aperture settings (including the possibility to set the lens aperture opening and closing limits).
![image 17.png](../../../__attachments/5018158023/image%2017.png?inst-v=5e52fa89-af48-462a-9bad-31c36e430795)
Camera depth of field
![image 17 (1).png](../../../__attachments/5018158023/image%2017%20(1).png?inst-v=5e52fa89-af48-462a-9bad-31c36e430795)
Lenses with various camera aperture settings
Illumination
Illumination should be 150 lux or higher. The lighting conditions within the camera control area should be ample and uniform, excluding any side light sources. For example, to control the entrance group, prevent the camera from being illuminated by streetlights.
Ensure uniform face lighting, a neutral background, and the absence of a side light source. The ISO/IEC 19794-5:2013 standard specifies detailed requirements for images.
Pixel density
Pixel density should be no less than 500 per meter within an entire anticipated recognition area. This requirement determines the camera resolution and the camera vision angle.
The face recognition conditions
To ensure accurate face recognition, the following requirements should be met:
The face exposure to the camera is 1.5 seconds and more.
The eyes are open and clearly visible.
No headwear or dark glasses.
The face image is sharp and clear.
No dark shadows.
The face proportions shouldn’t be distorted.