I'm performing some experiments with fish, and I'd like to analyse the videos with a software for animal tracking that automatically recognizes the animal and plots the track, makes calculations, etc.
In my tanks the fish are clearly darker than the background, so I think the software will recognise them. However, I've set a roof in one part of the tank as a shelter. My concern and my question is if softwares for animal tracking perform well when the focal animal (in this case the fish) appears and disappears from the image (because it is in under the roof) several times during the test.
A picture of the tank is attached.
David, it depends on the type of software. You will be on the safe side if ask developers of a software what their program would do if an animal temporarily disappears from view.
This question raises more questions - one thing to consider is whether you need to recognize specific fish or just track any given fish as noted between appearance and disappearance. I'm assuming that you just want to anonymously track between appearance and disappearance, otherwise I suppose you could mark or tag them, but not sure that's practical.
Hi David,
Much depends on the fish that you are using (size and color) and on the software that you choose.
Based on my experience I would say that your setup is ok except the wooden stick that crosses from side to side of the arena. Not being a shelter in itself, that could mess up quite a lot with the software. Is that some sort of gate that you open and close? If yes I would suggest a different strategy like confining your fish to a small clear container for acclimation. Something that at the beginning of your trial you can manually open to release the fish and remove from the arena.
Trust me, as Bart said, if you get the set up right first it will save you a lot of time down the track!
In our lab we use Ethovision, it is a very sophisticated animal tracking software. It is pricy but you can download and try the demo before buying it. http://info.noldus.com/free-ethovision-xt-trial
I highly recommend you to try the demo of this or other software before running you whole experiment. It takes a bit of trial and error before getting this kind of thing right.
(I am not affiliated to Noldus in any way).
Good luck and let us know how it goes!
Tullio
Dear David,
in our lab we use EthoVision as well, and (although I have never used it myself) there you can define hidden areas like shelters so that the software will know your subject is not lost but hidden. In my experience, lighting is everything in tracking. The shadow at the upper wall might be problematic, not only for the tracking but also for the animals' behaviour-they might try to hide there. So if you want them to consider all the tank (except for the shelter) as unprotected open space you should move either the lighting or the tank (whichever is more practical). Whether the wooden bar will cause problems might also depend on how large your fish are in relation, i.e. whether they will dissapear completely (no problem if you define it as hidden area), only a small part of the body disappears (at least in EthoVision you can merge areas detected as potential subject as long as the brak is not too big), or like half of the animal is invisible (which might be problematic indeed). What came to my mind as an additional potential problem is reflections of the water surface. This will of course depend on how much the surface will move, so if your fish stay close to the ground it might be ok. We did experiments with pigs on a black floor, which was supposed to be much lighter than our animals, but due to some light reflections our animals got lost. We changed the lighting (coverd the headlight with cloth to make it more indirect) and now it works.
Like Bart and Tullio, I strongly recommend intensive tests of the setup before you start, as most of the problems we had in the beginning could not be solved within the software but by changing lighting, wall paint etc.
Good luck,
Sandra
Thanks for all your comments. I have made some improvements with lighting, so although there are still some shadows in the wall (which are impossible to remove without creating reflections...), there are not so hard, and the fish contrasts quite well. I also removed the stick so there is no discontinuity between both parts of the tanks. There are no reflections right now.
I'll let you know how it goes!.
Thanks again,
David
I also use Ethovision solution and I build my observation tanks with a plexiglas floor under which I insert an infrared floor (available from Noldus but also very easy to build yourself) which eliminate all problems with lighting and allows filming under night conditions. Further, your top 'on the hidden' zone could be made of dark grey plexiglas which is transparent to IR and then you could also film your fish in the shelter and make different zones for analyses etc... Good luck, Marie-Laure
- Badges
- Science topic
- Similar topics
- Biological Science
- Zoology
- Ichthyology
- Vertebrates
- Fish