|
ANY-maze Help > The ANY-maze reference > The Protocol page > The elements of a protocol > Tracking > Animals Animals
OverviewANY-maze detects animals by looking at the contrast between them and the background of the apparatus. Thus, for ANY-maze to track successfully, the animals must be either lighter or darker than the background (or both). That said, even a very small level of contrast is usually sufficient for ANY-maze to track well - see figure 1. Just knowing whether the animals are lighter or darker than the background is sufficient for ANY-maze to track the animal's centre, but to track the head and tail, ANY-maze analyses the shape of the animals - which of course depends on the type of animal; a quail has quite a different shape to a mouse! You must therefore specify both the animals' type and their 'colour' in order to track them. You do this using the options on the Animals element of the protocol.
Figure 1. There's very little contrast between this white mouse and the sawdust in the Y-maze, but by telling ANY-maze that the animal is lighter than the background the system is still able to track accurately.
When your protocol includes a hidden zone, you will also need to specify the animal's approximate length.
Specifying the type of animalAs mentioned above, ANY-maze needs to know the type of animal in order to track the head and tail; it also uses this information to improve the performance of the Automatic starting of tests. If you're not interested in the head or tail position and you are not using auto-start, then it actually makes no difference what animal type you choose. Note that tracking of the centre point works for all animal types.
Mice or rats (other than hooded rats)These are the classic settings for tracking the animals head and tail. You will also need to specify whether the animals are lighter or darker than the apparatus background. The animals are hooded rats
This option is for tracking so called hooded rats, such as Long Evans or Lister hooded.
Figure 2. To track Long Evans rats you should select the option 'The animals are hooded rats'
When tracking hooded rats ANY-maze will primarily track the dark part of the animal (as this includes the head), although the tracking works differently to tracking an entirely dark animal (such as a black mouse), so it is important that you do choose the 'hooded rat' option. When tracking hooded rats, especially if you are tracking the animal's head, the tracking sensitivity can be very helpful. For example, if the animals are in a relatively dark arena, or one in which the light is uneven and they animal tends to throw a shadow, then decreasing the sensitivity can help. On the other hand, if the background includes some darker areas then increasing the sensitivity can help Quail
This option is specifically designed to track the heads of quails, it may not work well with other birds. Also, it assumes the animals are on a background that is lighter than the animals. Some other animalIf you are not tracking rats, mice or quail, then this is the option to select. Using it will track the animal's centre point correctly, but the head and tail tracking may not be accurate. As mentioned above, this is because the head tracking analyses the shape of the animal and unless the animal is a similar shape to a rat, mouse or quail, then it is unlikely to work.
Specifying the animal's colourNormally, it is easy to specify whether the animals are lighter or darker than the background, but there are a number of situations which require special attention.
There's very little difference between the animals and the backgroundIn this situation, if the animals are light but are very similar in colour to the background (for example, a white mouse walking on sawdust), then you should specify that the animals are lighter than the background. You may notice that there's a shadow around the animal, and so you could specify that they're darker than the background - however, tracking shadows is generally a bad idea and will give poor results. If the animals are dark but are very similar to the colour of the background, then you may need to experiment to see which setting gives the best tracking. Start by specifying that they're darker, and if this doesn't work very well then try the selecting the option The background and/or the animals are not uniform. The animals are white and the background is sawdust/wood-shavings
Tracking white animals on sawdust or wood-shavings is usually quite challenging, because in a black and white image the animals are often a similar colour to the background and while ANY-maze will often track the animals' centre in these conditions, getting accurate head tracking is often impossible . However, while in a black and white image the animal and background are similar shades of grey, in a colour image the animal and background are usually different, with the background being a shade of pale brown. In fact, some 'white' rats are also a yellowy brown and even in a colour image they're quite similar to sawdust - but in most cases there's a more distinct difference between the animal and the background when viewed in colour. ANY-maze can take advantage of this difference by enhancing the animal-background contrast by using the colour information if it is available. At present this is only supported in colour videos, for example, mp4 or wmv videos, and it does not yet work with colour cameras. If you select this option then ANY-maze will analyse the colour data if it is available and enhance the animal-background contrast, note that the resulting image will still be black and white. If colour information is not available, then this option is identical to selecting the Animals are lighter than the background.
Figure 3. The image on the right comes from a colour video so ANY-maze has been able to use the colour information to enhance the contrast as shown in the image on the right. Note that the enhanced image is used internally, ANY-maze still displays the unenhanced image.
The animals are white and the background is pale grey
Tracking white animals on a pale grey background is usually challenging, because in a black and white image the animals will appear to be pale grey and therefore present no contrast to the background. However, while in a black and white image the animal and background are similar shades of grey, in a colour image the animal will usually appear a pale yellowy brown and ANY-maze can use this colour information to, at least, detect the centre of the animal. Note that, at present, this is only supported in colour videos, for example, mp4 or wmv videos, and it does not yet work with colour cameras. If you select this option then ANY-maze will analyse the colour data if it is available. If colour information is not available, then this option is identical to selecting the Animals are lighter than the background.
Figure 4. The image on the left comes from a colour video, but ANY-maze sees the image on the right, in black and white, where there is essentially no difference between the colour of the animal and the background. However, by analysing the colour data ANY-maze is able to track the animal's centre.
The animals are not a uniform colour (and are not hooded rats)If you wish to track non-uniform animals (other than hooded rats - for these, see the previous section) then obviously they won't be either lighter or darker than the apparatus background. In this case, you can do one of three things:
The background is not a uniform colourThis situation is similar to the one described above (where the animals are not uniform), but in this case you can't use either of the Animals are lighter or the Animals are darker options because if you do, the system will only track the animal in some parts of the apparatus! The only choice in this situation is to choose the option for The background and/or the animals are not uniform. See figure 5 for an example of when this option is required.
Figure 5. In this hole board apparatus, the animal is lighter than the background in the centre of the apparatus but is darker than the background around the edge. However, by using the option for a non-uniform background, ANY-maze tracks very well.
Note that tracking hooded rats on a non-uniform background can be problematic if the background included quite dark areas. In this case you will probably need to increase the tracking sensitivity. Some animals are lighter and some are darker than the backgroundIf you're working with different strains of animals, then some may be lighter and some darker than the background of the apparatus. In this case, you should specify the option Some are lighter and some are darker and then choose whether ANY-maze should automatically determine the animal's colour or whether you will enter it manually.
Specifying the animal's lengthIn some circumstances, you will need to specify the animals' length:
The reason ANY-maze needs to know the animal length is so that it can determine when the animal becomes visible if it is hidden at the very start of the test - otherwise, it would have no idea how large a target it should be looking for.
In either of the above circumstances you will be asked whether all the animals are approximately the same length or whether they have different lengths. If all the animals are broadly the same size (which often means the same age) then you should say they are approximately the same length, as the small variations in size between individuals won't matter. However, if your experiment will include animals of different species, strains or ages that are of significantly different sizes, then you should say they are not the same length; in this case you will be able to enter the length of each animal individually on the animal's spreadsheet. If you specify that the animals are approximately the same length, then you will need to enter that length including units of m (metres), cm (centimetres) or mm (millimetres). The length you enter can be quite approximate, but should exclude the animal's tail. For mice, a value of around 6cm is usually about right, and for rats around 18cm - although this will depend on the age and strain of the animals. Note that when the protocol includes a hidden zone, the length you enter is used when a test is being run, therefore changing it after a test has been performed will not alter how ANY-maze scores the zone exits. However, in the case where the length was required because the protocol is only detecting freezing, then, in common with almost everything else in the protocol, changes made after a test has been performed will cause the tests' results to be updated. Adjusting the contrast sensitivity
Imagine you have a 'white' rat on a brilliant white background - in this case, the animal will almost certainly be darker than the background and selecting the option The animal is darker than the apparatus background will work correctly. However, imagine that the animal now has a black skull cap fitted - this will offer very high contrast to the background, whereas the animal's body only offers low contrast. In this case, ANY-maze will tend to consider the high-contrast skull cap as the target to track and will ignore the low-contrast body of the animal - clearly, this is not what you want. To overcome this, you can increase the Contrast sensitivity so that ANY-maze considers a wider range of contrasts as being part of the animal.
© Copyright 2003-2026 Stoelting Co. All rights reserved ANY-maze help topic T0312 |