IQinVision IQanalytics - Manual (Page 9)

9
5 Camera Placement & Guidelines
This section provides recommended camera angles and positions.
Camera Positioning
Camera positioning is critical for capturing good data for video analytics. Care should
be taken to avoid a) object size being set too large, and b) occlusions.
Typically, cameras should be installed with one of three views:
1. Top-Down view (vertical ceiling mount)
2. Perspective or Angled view (wall or corner mount looking down) – Useful for
detecting general intrusion.
3. Side View (horizontal wall mount at lower height).
The Top-Down View is generally good for counting, detecting direction of movement,
etc. The second option (perspective or angle view) is good for detecting intrusions.
The third option (side view) should only be used for face detection. It is not
recommended for other analytics since objects can be hidden from view (occlusions).
FOV (Field of View) and Object Size
For people detection, the object or person size should be 10-20% of the whole scene.
The object size should not be set to occupy 40-100% of the scene.
Recommended Scene and Lighting Conditions
General recommendations for lighting when using video analytics.
1. The scene should be well-lit. Low lighting levels produce video noise that can
triggers false object alerts.
2. Stable and consistent lighting will provide good results. Fluctuations in lighting
or uneven lighting at the scene will cause false alarms in any video analytics
system.
3. If providing appropriate lighting is an issue, it is recommended that a camera
with IR illumination be used.
4. Avoid occlusions (one object going in front of another). This will break object
tracks and will cause object disappearance and other related problems.
5. Avoid pointing the camera directly at light sources as described below. The
following situations will make the video unusable for analytics and will
temporarily signal saturation.
a. Camera pointed directly at the sun.
b. Camera pointed directly at light source
c. Camera looking at vehicles with headlights (cars, trains in tunnels)
6. Avoid or minimize shadows by repositioning the camera or light source(s).
7. If possible avoid the presence of trees and other vegetation in the camera’s field
of view. Swaying or movement of trees and bushes in the wind can cause false
alarms. If they cannot be avoided, use ‘ignore area’ – these zones will be ignored
during processing.
8. If possible, avoid the presence of water areas – rivers, lakes etc. in the camera’s
field of view. Water areas can cause reflections and ripples that can cause false
alarms. If they cannot be avoided, use ‘ignore area’ – these zones will be ignored
during processing.