Documentation for Face Intellect 7.1. Documentation for other versions of Intellect is available too.
The use of high-resolution video surveillance cameras is recommended for the correct performance of Face Intellect software package. Video surveillance cameras used with Face Intellect software package must support color video.
For the proper face recognition using Face Intellect software package it’s necessary to install and configure the video cameras in such a way that the following requirements are fulfilled:
- The frame rate must not be less than 12 fps and not less than 6 fps for face recognition at turnstiles.
- The maximum deviation of the face in the camera FoV from the face-forward position should not exceed 15 degrees.
- The distance between pupils on the received photos of the captured face must correspond to the requirements of the used recognition module:
- Cognitec: not less than 96 pixels;
- Tevian: not less than 32 pixels;
- VisionLabs: not less than 50 pixels.
- There should be minimum mutual overshadowing of the captured faces.
- The faces should be evenly illuminated with a diffused light of at least 200 lux. Directional side lighting is not allowed.
- The sharpness of the captured faces on the received photos must be over 64 grayscale. Deficient or exceeding illumination is not allowed.
- There should be no back light and sharp gradients of light and shade.
- The photos of the captured faces received from video cameras should be clear. Image blurring caused by motion is not allowed.
The required distance between the camera and the face can be set using a lens with required focal length.
For VisionLabs recognition module the size of the photo added to the database is configured in the CascadeDetector::SizeHint settings section of the config.xml file located in the <Intellect installation directory>\Modules\FaceRecognition\VisionLabs\data.