Aim
To choose which technique EthoVision XT uses to detect the nose-point and the tail-base points of your subjects.
Prerequisites
▪In the Experiment Settings, under Tracked Features, you selected Center-point, nose-point and tail-base detection.
▪The camera must be placed overhead.
Procedure
1.Choose Setup > Experiment Settings.
2.Under Body Point Detection Technique, choose the option that applies to your experiment.
Contour-based
This technique uses foreground/background segmentation based on conventional image processing. The center-point, the nose-point and the tail-base point are derived from the detected body contour. This technique is suitable for any animal with an elongated body and pointed snout, for example rodents, fish, and certain insects.
note Contour-based is used by default in older EthoVision XT versions.
If you choose this option, the nose-point and the tail-base point are detected using a trained neural network. The center-point is detected based on the contour of the detected subject, as described above (except for when there are two subjects in the arena; in that case the center point is also found using the neural network).
important Deep learning is currently limited to rodents, and for up to two subjects. If you want to use the Deep learning technique, consider that there are specific restrictions and hardware requirements. See also:
Notes
▪If you chose Center-point as the tracking feature, the center-point is always detected with the Contour-based technique.
▪You cannot use the Deep learning technique combined with Behavior Recognition.
▪When you select Deep learning, you may see this message: The graphics card driver is outdated or not installed or the graphics card does not support CUDA. It is possible that the graphics card is too old, or the graphics card driver is not up-to-date. See Deep learning: Requirements