In addition to the ideal field of view, camera placement, and floor-plane grid, you can also control the direction of detection.

Field of View

The camera must be able to see the feet of people as they pass through the doorway. The ideal Field of View (FoV) for the camera looks something like this:

Please read this article on defining the floor plane 3D Grid.

Any camera connected to Camio can include a 3D grid that defines the layout of the space in three dimensions in the camera's Field of View (FoV). Within the grid, specific tiles are designated as the transition between inside and outside of the doorway.

With the grid in place, Camio identifies the number of unique individuals entering through the door at any point and, in the case where the number doesn’t match the number of authorized credentials presented to the access control reader, detects the potential tailgating event.

These are the recommendations to help Camio perform the most accurate tailgating detections.

Camera Placement

The most important element in this setup is the placement of the camera such that it best captures people entering through the door. 

• The camera should have a complete view of the door.
• The camera's Field of View (FoV) should extend to the floor such that it captures the person's feet as they touch the floor.
• The camera should not be centered facing the door and instead should view the door at a certain horizontal angle so that larger people in the foreground cannot completely occlude smaller people behind them.
• The horizontal angle should not be too steep.
• The camera should be looking downwards at the door with a relatively wide angle to capture more of the floor.
• The vertical angle should not be so steep as to exclude the heads of people passing through the doorway.
• The door should not swing open such that it blocks the view of people passing through it.

 

Sample cases describing setup scenarios

		Vertical angle is not too steep. The door is at a reasonable horizontal angle from the center. Floor and door are covered in full detail. Some information of the area outside the door will be available as the door opens.
		Camera is placed right in front of the door, so individuals in the foreground can completely occlude those in the background.
		The horizontal angle of the camera is too steep and it will not get sufficient information of the moment someone enters through the door. The camera is only providing information of the area inside the room and not sufficient context of the entrance.
		A significant portion of the floor is missing from view, so Camio can't see the person's first step after entering into the room. Without the view of direct contact between the feet and the floor, Camio doesn't get precise information about the person's location in the scene.
		The camera's view of people passing through the doorway gets blocked by the door as it opens.
		The vertical angle for the camera is too steep With a steep view, the individuals are more head focused and the area on the floor for detection is reduced.

 

Drawing the Grid

The 3D grid helps Camio understand the transition between inside and outside the doorway. Some key optimizations while drawing the grid are as follows:

• The grid should be sized sufficiently large enough to cover the doorway and surrounding floor.
• The grid should not be too large or too small; cover only the area clearly associated with entries and exits from the room. 
• An ideal grid will be large enough to allow for a row above and below the outside and inside grids for better detection and analytics.
• The 'outside' and 'inside' tiles on the grid act as key indicators of the areas on the floor that are relevant to tailgating detection.
• The 'outside' tile should cover a small portion of the inside of the room so that a detection through an opaque wall or an opaque door gets some visual context of people entering into the room.
• The 'outside' tile should not extend more than a foot inside the room in order to prevent individuals from stepping on both 'inside' and 'outside' tiles even without leaving the room.
• More than one tile can be marked as 'inside' or 'outside' in order to capture the necessary points on the floor.
• There can be empty tiles in between 'outside' and 'inside' tiles, as long as the 'inside' tiles cover all the directions from which the person can enter the room.
• Empty tiles in between 'outside' and 'inside' tiles often help reduce false positives as they provide a buffer against transient errors.
• Important: Even in the case of glass doors, if the stiles or the rails are opaque, they might partially block the view of the people behind the door. Due to this, it is recommended to extend the 'outside' tile 1-2 feet inside the threshold, in order to ensure detection of the person while entering inside the room.

 

Sample cases describing good and bad grid layouts

		A small portion of the 'outside' tile lies inside the room, providing sufficient context for the individual entering in. The 'inside' till is completely inside the room. There is a row of tiles outside of the door to provide more time for the analysis There is a safe gap in between 'inside' and 'outside' tiles to prevent transient errors.
		A small portion of the 'outside' tile lies inside the room, providing sufficient context for the individual entering in. The 'inside' till is completely inside the room and covers the region that a new person entering the room will always interact with. Would be more ideal to have an extra row outside of the door and a gap.
		The 'outside' tile is almost completely outside the room and may not sufficiently capture a person entering when the door is opaque.
		The entire grid is entirely inside the room, so a  person walking inside the room without passing through the doorway can still step on both inside and outside tiles to create a false signal.

 

Direction of Detection

By default, Camio counts people in both directions outside => inside and inside => outside. You can change that default on the integration page for your access control system by unchecking the direction you want to exclude from the count under Tailgating Detection Options.

 

---

Learn more in the article about receiving tailgating alerts.