Using ImageJ to automate nest counts


#1

I have used ImageJ threshold to count flocks of birds before. I was hoping to do the same with nest counts. One of my biggest problems is the eggs are often the same color as the substrate, this are ground nesting beach birds. Is there another way other than threshold to get to the same goal?

The attached photo will give you an idea of how well this eggs blend in to the background.

Sorry I am new to this program and I’m hope someone can point me in the right direction.

Thanks,
Zak


#2

Good day!

I have used ImageJ threshold to count flocks of birds before.

Sorry I am new to this program […]

??

[…] how well this eggs blend in to the background.

I think this was nature’s idea …

What about the temperature of the eggs, i.e. how do infrared images look like?

Regards

Herbie


#3

I see you have question but ?? doesn’t help me to explain in more detail, because I do not know what you are asking.

Yes that is nature’s idea. But I wanted to give the group prospective.

I have you used ImageJ to count flocks of birds by using the threshold tool to isolate the darker patterns of birds and then ran analyze particles to give me a count. But since there isn’t much contrast between the eggs and the substrate I am wondering if there is a different tool that is better suited for the task. There is a possibility that temperature may show the nest but not always. Plus I do not have a thermal camera. More then likely the eggs will show cooler then the surround area, since this will take place in the middle of summer. But the colony might not all be incubating at them same time, making those nest the same temp.

The attached image is of wooden eggs that are the same size/shape and coloration as the intended target and a decoy. So they will show no heat.

Thanks,
Zak


#4

I see you have question but ?? doesn’t help me […]

It’s just to point out that you appear not so new to the program.

Because I’ve difficulties finding the eggs by eye, machine vision won’t help you, except you are willing to invest in advanced pattern recognition.

Regards

Herbie


#5

Ha, well yes I have used some of the features. So I am not completely green. But I am talking to the unknown world of ImageJ forum users and I’m sure there is a wealth of knowledge that is far beyond my short term usage.

I agree with your conclusion, but I wanted to see if anyone else had ideas. It never hurts to hope and dream right?

I will add that ImageJ is still useful even with out the semi-automated counts by using the grid system and multi-point feature to mark the nest that you can visually see. I also do the same in ArcMap that way I have a feature class with lat, long as well.

Thanks!
Zak


#6

Hi Zak,
The nests seem to be circles with subtly different Hue than the surrounding and void of debris; if more than one egg is present, there seems to be a darker shadow line separating them, shape and brightness being absent elsewhere. Maybe that helps.
As Herbie suggested, can you obtain images recorded at other wavelengths than visible and do they perhaps show the eggs better? This article (just a google search result, no interest or other purpose) seems to suggest you may be lucky:
https://openi.nlm.nih.gov/detailedresult.php?img=PMC3398933_pone.0040211.g002&req=4