Methods for estimating the body points

Depending on the experiment mode (standard vs. DanioVision) and how many subjects and body points you specify in the Experiment Settings, the Method section of the Advanced Detection Settings window lists one or more options.

 

note  In some cases, like when tracking the center point of one subject, this group of options is not available, because EthoVision XT applies the only method possible for that experiment.

Any Species / Default

▪This was called Shape-based / Any in EthoVision XT 14 and earlier.

▪This method analyzes the contour of the blob detected as subject at each sample to assign the nose-point and the tail-base. If you track rodents, make sure that the tail is fully detected.

▪The subject’s center-point is calculated as the geometric center of the area detected as subject. See Calculation of the center-point below for details.

▪This method is automatically activated when you track the center-point in one animal per arena, and you specified Other species in the Automated setup.

▪If your experiment is from an older EthoVision XT version, you have additional options available. Here you can choose to use the last improvements or keep the settings as they were in the older experiments.

▪Choose Priority on tail base position whenever the tail is apparent in the video and well detected. This method finds the tail base and then assigns the nose-point to the opposite point in the subject contour. If the tail base is not found, then other rules based on the contour are used to find the nose-point. Choose Balanced (XT 14 and earlier) if you want to use the same method as in previous EthoVision XT versions, or if for any reason the first option does not give good results.

Rodents / Default

▪This was called Simple Model / Rodents in EthoVision XT 14 and earlier.

▪This method analyzes the shape of the blob detected as subject and builds up a “rodent model”. It is more robust than the Any species / Default method because it does not require the nose and tail to be visible: it can 'predict' the position of the nose and the tail based on previous samples.

▪The subject’s center-point is calculated as the geometric center of the area detected as subject. See Calculation of the center-point below.

▪Choose this method when you want to track a single rodent without occlusions or without difficult tracking conditions, like in an open field or a novel object recognition test. For optimal results, make sure in the detection settings that the tail is removed from the body contour, using Erosion and then Dilation. See Advanced detection settings: Subject contour

▪If you upgrade your experiment from an older version of EthoVision XT, select Extra correction of nose-tail swaps to use the latest algorithm that aims at correcting nose-tail swaps during tracking. Compared to previous versions, this algorithm corrects swaps faster, reducing the total number of samples with swaps (if they occur). Choose EthoVision XT 15 and earlier if you want to keep using the old algorithm.

 

The extra correction is based on the detection of false backward movements, that is, when the animal looks like it is moving backward when the nose and the tail-base points are swapped.

Rodents / For occlusions

▪This was called Advanced Model / Rodents in EthoVision XT 14 and earlier.

▪This method learns the animal shape and how it moves in the first 15 frames and continually updates its statistics. Therefore, it can handle severe shape distortions, such as, for example, when the subject’s body is occluded or when two subjects come into contact. However, it requires a lot of computer performance.

▪The subject’s center-point is calculated as the geometric center of the area detected as subject. For experiments created with previous EthoVision XT versions, choose how EthoVision XT should find the center point. See Calculation of the center-point below for details.

▪Choose this method for rodents in a social interaction context, or when you track rodents that can be occluded, for example, by bars or other objects in the cage. For optimal results, make sure in the detection settings that the tail is removed from the body contour, using Erosion and then Dilation. See Advanced detection settings: Subject contour and Subject contour for nose-tail base detection

Adult Fish / For occlusions

▪This was called Advanced Model / Adult Fish in EthoVision XT 14 and earlier.

▪Choose this method to track adult zebrafish and other fish of similar shape, when viewed from above. Make sure that the whole body of the fish is detected, including the tail. This method is recommended no matter how many fish you track in the same tank.

▪The subject’s center-point is calculated as the geometric center of the area detected as subject. For experiments created with previous EthoVision XT versions, choose how EthoVision XT should find the center point. See Calculation of the center-point below for details.

Other species / For occlusions

▪This was called Advanced Model / Other in EthoVision XT 14 and earlier.

▪This method does not make any assumption about the shape of the subject. Note: with this method you only track the center point.

▪The subject’s center-point is calculated as the geometric center of the area detected as subject. For experiments created with previous EthoVision XT versions, choose how EthoVision XT should find the center point. See Calculation of the center-point below for details.

▪Choose this method to track for example insects, crustaceans or large animals, and when multiple subjects move in the same arena, or when occlusions make detection difficult.

▪Also choose use this method to track fish from the front view of the water tank, and the Adult Fish method does not give good results.

DanioVision

▪This method is optimized to track zebrafish larvae on a white background. It is only available in DanioVision experiments.

▪Moving the Sensitivity slider to the left results in detecting more of the body of the larvae, but also more noise. Moving the slider to the right results in detecting less noise, but also gives less good detection of the larvae.

▪For more information, see the DanioVision DVOC-0041 - Reference Manual.

Method names

Please note that the names of the methods above differ from those in EthoVision XT 14 and earlier versions:

 

Calculation of the center-point

For new experiments

In all new experiments, the subject’s center-point is determined using the area of the blob (in yellow, per default) that indicates detected subject. The x, y coordinates of the center-point are the average of the x, y coordinates of each pixel in the blob. For example:

 

If the tail is detected, that contributes to the position of the center-point.

Experiments created in EthoVision XT 14 and earlier

If your experiment was created in a previous EthoVision XT version, and you use one of the For occlusions methods, you can choose whether to use the detected area (as described above) or the shape model. Choose Use center of model (XT14 and earlier).

The x, y coordinates of the center-point is the average of the pixels coordinates of the model fit over the detected subject. In many situations the model has a more regular shape than the detected area. For example:

 

Choose this option Use center of model (XT14 and earlier) to keep compatibility with older experiments. Consider that, if detection or the model parameters are not optimal, the model may differ greatly from the subject’s image. In this case the center point may shift compared to the expected position (see B vs. C in the next picture). This may add up distance moved on the long run.

Below: A. Subject not detected. B. Center point determined as the model’s center. C. Center point determined as the center of the detected area.

 

If that happens often, choose Use center of detected area. Or, for multi-subject tracking, adjust the modeled subject size. See Advanced detection settings: Subject size (multiple animals per arena)

See also

▪Adjust the settings for nose-tail base detection (Contour-based)

▪Subject contour for nose-tail base detection