Using the camera at night

This document describes potential issues when using Object Analytics at night.

Night mode affects video quality

The Axis camera's night mode is extremely good and achieved using their Lightfinder technology and IR sensor data from the camera. Still, even with this technology the resulting video has reduced contrast compared to daytime footage, since the video stream is only in grayscale. The lack of contrast may result in reduced person recognition in certain cases.

Night mode can also, in some cases, generate “grainy noise” in the video which can cause problems for motion detection. This is described in the article Reducing video noise when using IR-mode (night mode).

Triggering a scenario mask at night

As mentioned in the section above, the video stream at night is gray scaled. This means that the camera will have slightly reduced capability to detect a person based on the AI model, compared to fully colored daytime footage. The scenario mask might therefore need to take this into account if the placement is not optimal.

Entering bed in total dark room

Detecting a person at night might be a problem if the person is being partly covered by an object (here; the bed). The example below shows a person entering the bed from the far side of the bed and where the person is only partially visible.

Entering the bed in well lit room

In a well lit (illuminated) room, the person is detected even if partly covered since the video quality is higher than when using IR mode (gray scaled).