How and where you set up your cameras affects the performance of people counting analytics. An optimal camera setup can minimize missed or false counts.
These best practices apply to People counting and Tailgating detection scenarios that track people crossing a line in the camera's field of view. For crowd estimation, refer to Best practices for crowd estimation.
- People can be seen traveling along the longest side of the image.
- The camera has a top-down view (90° from ceiling).NOTE: Do not use fish-eye cameras or cameras with high distortion. The horizontal field of view should not exceed 120°.
- There is no motion blur in the video. Faster shutter speeds produce less motion
- The image is high-contrast, but not under-exposed or over-exposed.
- The frame rate is 12 fps.
- The resolution is 720p.
- People do not linger where the counting line is, such as security guards at store entrances or groups of people waiting in queues.
- Each entrance is monitored by one camera. If more than one camera is required to monitor a large entrance, delimit the cameras' view of the entrance with furniture.
- The area you want to monitor is well lit.
- The background is uniform.Example: Highly patterned floors might decrease the accuracy of detections.
- People crossing the line can be fully seen in the video.Example: The heads of the people crossing the line must be in frame.
- The entrance that you want to monitor is centered in the image. The counting line
configured in the analytics scenario should be drawn where the entrance is.