search

gilestrolab / ethoscope

a platform from monitoring animal behaviour in real time from a raspberry pi
http://lab.gilest.ro/ethoscope/
GNU General Public License v3.0
17 stars 25 forks source link

How to track very slow or immobile animals ? #195

Open augustinhebert opened 5 months ago

augustinhebert commented 5 months ago

Hi !

I am using ethoscope to track ticks. They are very slow animals and often remain in quiescence (ie not moving at all) for extended periods of time.

I noticed that ticks that don't move since the beginning of the experiment are not detected by the tracking algorithm. Those that were detected but stopped for a long time were "lost" after some time.

I tried to take a look in the adpative_bg_tracker.py file but I am not well educated in Python and I struggle to understand the code. Can you suggest anything that I change or add in this file order to detect my immobile ticks ?

Thank you in advance !

Augustin Hebert

ggilestro commented 5 months ago

Hi Augustin,

The current tracking procedure uses background subtraction, meaning that if the animal is not moving the system will assume it's just part of the background and will not look for it. There are other possible solutions one could use but they are not coded in. The easiest option for now would be to use the ethoscope to record videos, perhaps with slow FPS to save bandwidth, and the use a software like SLEAP.AI or idtracker.AI to track the animals offline.

Other solutions require coding. They are not particularly difficult and could be fully implemented in a couple of weeks of work including testing. At the moment I don't have anybody in the lab who could help with that, sorry.

augustinhebert commented 4 months ago

Hi Giorgio,

Thank you for the information, unfortunately our computers aren't powerful enough to run this kind of softwares, but they seem like a great tool !

No worries for the coding, our solution is to assume that undetected ticks stay in the same position, and we can find this position using the img snapshots from the db file. It takes a bit more time but that way we can fill the gaps in our tracking data.