colour selective trap.


I've been asked to help someone who's working on a program to re-introduce red squirrels into areas of the UK.

I'm looking at the feasibility of designing a squirrel trap that will trap and hold grey squirrels, but release red squirrels. To do this I need to be able to detect the colour of the animal entering the trap.

Sparkfun sell several colour sensors....

My current idea is to mount a colour sensor on the side of the entrance hole - as the animal enters the trap the sensor gets triggered and if it detects red fur nothing happens, but if it detects grey then the door snaps shut.

Has anyone any experience of using any of the above sensors - they seem to vary greatly in price, but what about performance? how sensitive are they to ambient light? will I get similar performance at dawn and dusk as well as midday? I'm not looking for 100% accuracy, the idea is that the trap would be checked frequently, so even if a red one gets shut in by mistake it can be released later - but I am looking for at least 80% accuracy.

Any suggestions? - particularly from anyone who's used any of these sensors.


This is a difficult one, the squirrel would probably have to be lit so the colour sensor can get an accurate reading. Another issue would be the accuracy even then and the red squirrel’s winter coat (which I believe is more grey).

In my experience (although I have none with colour sensors) detecting moving things with light is a pain.

Are there any other major differences between the squirrels that can be detected, aside from colour?

With a colour sensor you could measure the colour as the squirrel moves and take an average colour, maybe if you set the trap out for a while and capture some data (without trapping things) while watching yourself to see which squirrel type passes through /hopefully/ you’ll see some difference in the readings. That could be a bit of a long process though if your squirrel population is similar to mine (I’ve been watching a tree for 2 minutes and no squirrels at all).

Sounds like a good project, I’m looking forward to seeing some red squirrels in my garden soon :stuck_out_tongue:

If you haven't seen it yet, squirting squirrels has been done before. He talks about color discrimination around the 10 minute mark. He does use an Arduino, for aiming the turret and pulling the trigger, but needs more processing power for blob classification.

If you haven’t seen it yet, squirting squirrels has been done before

I’ve just watched it, he’s talking about something slightly different, he’s trying to do image recognition on an object many meters away. My intention is to have a colour sensor right next to the squirrel as it enters the trap - I only want to differentiate between red and grey - hopefully a far easier task! - although the water cannon idea does look good!