Skip to main content

Camera Positioning & Counting Line Guide

Practical fundamentals for accurate people counting

This section explains how to position a camera so that people can be tracked reliably across a counting line.
It applies to both fisheye (top-down) and angled CCTV installations.

The goal is simple:
Maximise usable tracking area and minimise occlusion.

1. Coverage Area vs Usable Tracking Area

Recommended image:
Top-down fisheye view divided into center (core), mid, and outer edge zones.

Not all visible areas are equally reliable for tracking.

  • The center area of the image provides the most stable tracking
  • Areas closer to the edge are more distorted and less reliable
  • Extreme edges should be avoided for counting lines

Guideline
Place the counting line inside the central usable area, not at the image edge.

2. Counting Line Placement

Recommended image:
Two examples side-by-side: counting line in center vs counting line at image edge.

A counting line works by tracking a person:

  • Before crossing the line
  • During the crossing
  • After crossing the line

If a person is only detected on one side, the crossing is incomplete and may not be counted.

Good practice

  • Leave visible tracking space on both sides of the counting line
  • Avoid placing the line at the boundary of camera coverage

3. Fisheye Camera Mounting (Top-Down)

Recommended image:
Ceiling-mounted fisheye above a doorway, with doorway inside the central circle.

For fisheye cameras:

  • The central circular area provides the best tracking quality
  • Tracking degrades gradually toward the outer edge

Good practice

  • Align the doorway or walking path inside the central area
  • Slightly off-centre mounting is acceptable if the counting line remains inside the core area
  • Keep the camera level to avoid shifting usable coverage

4. Person Size in the Image

Recommended image:
Three people examples: too small, correct size, too large.

Tracking reliability depends on how large a person appears in the image.

  • Too small → detection becomes unstable
  • Too large → people overlap and block each other

Good practice

  • Adjust mounting height and distance so full bodies are clearly visible
  • Avoid placing the counting line too far from the camera (people too small)
  • Avoid placing it too close (people too large)

5. Visual Blockage Check

Recommended image:
Camera view partially blocked by exit sign or hanging decoration.

People counting is a visual process. Any obstruction affects tracking.

Check for:

  • Exit signs
  • Hanging decorations
  • Lighting fixtures
  • Ceiling structures

Good practice

  • Ensure the counting line and surrounding tracking area are fully visible
  • Reposition the camera if any object blocks people near the line

6. Angled CCTV Camera Considerations

Recommended image:
Comparison of steep angle vs shallow horizontal CCTV view.

Angled cameras are more sensitive to positioning.

  • More horizontal angles increase body overlap
  • Occlusion increases as people block each other
  • Tracking reliability drops faster than with top-down views

Good practice

  • Use the steepest practical angle
  • Avoid placing the counting line far from the camera
  • Ensure people remain visible on both sides of the line

7. Maximising Coverage Without Reducing Accuracy

Recommended image:
Wide entrance covered by a single fisheye with clear central tracking zone.

When covering a wide area:

  • Prioritise usable tracking space over total visible area
  • Ensure the full walking path crosses the counting line within reliable coverage

Key point
A wider view is useful only if people can be tracked before and after the line crossing.

8. Final Installation Checklist

Before finalising installation, confirm:

  • Counting line is inside the usable tracking area
  • People are visible on both sides of the line
  • Person size is appropriate (not too small or too large)
  • No overhead or foreground blockage
  • Camera angle minimises occlusion

If these conditions are met, tracking accuracy will be stable.