Troubleshooting: Detection

Issues

I cannot modify the Detection Settings

The yellow blob changes its size periodically

The percentage of Missed samples is very high

The detected nose point is far from the animal’s contour

 

I cannot modify the Detection Settings

Once data has been acquired in EthoVision using a specific profile (Arena Settings, Trial Control Settings, or Detection Settings), they become locked from changes. This ensures that EthoVision always preserves the settings you used to acquire your data, so it can be checked or replicated.

If you need to make changes, right-click your existing settings profile, and select Duplicate. The new profile will allow modifications.

The yellow blob changes its size periodically

The area considered as detected subject changes at regular intervals, as if the brightness of the image was not constant.

inset_3201204.jpg 

Check that the auto iris function of the camera is not enabled.

For digital cameras, increase the Exposure time. It may be that periodical changes in the intensity of the ambient light are picked by the camera sensor.

The percentage of Missed samples is very high

inset_3301205.jpg 

There may be a few reasons for this:

The video resolution of the camera and/or the frame rate are set too high for the current detection method. EthoVision XT cannot acquire so much video information and at the same time track the subject. Reduce the frame rate and if necessary the video resolution in the Video source.

The camera may not be sending the video frames at the expected rate. This occurs often with webcams or IP cameras. Select Dropped frames correction.

The detection method currently selected is using too much processor time and that may not be necessary. For example, try a less demanding method like Static Subtraction instead of Dynamic Subtraction or Differencing.

See also Sample rate

The detected nose point is far from the animal’s contour

This may happen in some cases with detection based on Deep learning. For example, the nose is placed somewhere on an object or on the image of the subject reflected on a wall, as shown in this figure:

inset_4101206.jpg 

In the Detection Settings, under Method, click Define and reduce the size of the box around the animal.

inset_4201207.jpg