To collect the most accurate data, use cameras that capture natural bottlenecks (aka choke points) where people in the store must pass through. Ideal camera placement should be nearest the choke point, which may differ depending on the use case.
Examples of choke points include:
Line Cross Choke Points = Entrance/Exits, Aisles, Sidewalks
Zone Tracking Choke Points = Vestibules, queue areas, waiting room
The further the camera placement is away from the choke point, the harder it will be for the AI to detect people.
See the following tables for examples of ideal camera placement.
1. Field of View for Occupancy Counting (Line Cross)
To get the most accurate data when using Line Cross, choose or position cameras to the height and angle as outlined below:
Height: Install the camera between 2.5 to 4 meters above the ground.
Angle: Tilt the camera downward at a 30-45 degree angle for optimal coverage.
In the example below, the camera is pointed outside of a vestibule where people will enter the store. This is an ideal place to draw your line because there is plenty of time to detect people as they enter the frame. Customers who enter through the door and cross the line will appear “big enough” to detect.
2. Field of View for Zone Tracking
To get the most accurate data for Zone Tracking, choose or position cameras to match the following guidelines:
Coverage: Ensure the camera's field of view encompasses the entire area where people pass.
Positioning: Center the camera to avoid blind spots.
In the first example below, the camera is pointing at the customer queue and there are no obstructions present, making this an ideal camera angle for Zone Tracking.
In the second example below, the camera is facing display shelves and is placed ideally to capture loitering by using Zone Tracking.
3. Lighting Conditions
For all AI features, ensure optimal lighting in the camera view as outlined below:
Avoid Glare: Position the camera away from direct sunlight or reflective surfaces.
Consistent Lighting: Ensure the area is evenly lit for accurate counting.
Optimal Lighting | Poor Lighting |
 Well-lit and easier for the AI to detect people. |
 The location is too dark. Poor lighting can degrade the AI’s ability to detect people. |
4. Avoid Obstructions
For all AI features, ensure obstructions in the camera view are minimal as outlined below:
Clear Path: Keep the camera's view free from obstacles like signs or decorations.
Regular Checks: Inspect the area periodically for new obstructions.
Obstruction Example 1 | Obstruction Example 2 |
 The entrance door is obstructed by multiple objects and aisles are also obstructed. |
 There are too many obstructions blocking the customer queue. |