Hello all, how are you? I'm a software engineering student and i'm currently working with some researchers and biologists trying to make a software to identify ants species (using a picture of their heads). Every specie has some particular characteristics, like the size of the eyes, the mouth and those things. Do you think that will be too difficult to do with this code?
I know this question is already 2 years old but how did you manage to do that?
If you wrote any blog post, if you have any resources to share, or better if you wrote a thesis, I would be interested to read it.
Thank you
Feel free to add me in G+ if you want to talk with me about the project. https://plus.google.com/+AleMoyano992
It's just a haar classifier. I believe there is already a cat.xml shipped with OpenCV. Learn about haar training.
If the ants look very similar except for some details, then this won't cut it however. Cascade model training is more for focussing on a certain class, which is ants in this case, to be able to locate them. Then you should diverse using other machine learning techniques.
Asked: 2014-04-01 19:22:02 -0600
Seen: 2,562 times
Last updated: Apr 05 '14
Hey. I'm doing something similar to that. Just curious, what college are you attending? I have some private questions for you. Let me know if you're interested in chatting.
could you clarify, if you need to distinguish several ant species, or if you need like ant=yes, cat=no
Is the picture provided always from one perspective (e.g., front or side), or will it be from any random perspective? That can make a difference in which detection approach you take.
I'm attending to Universidad Nacional del Sur in Argentina. Sure, i don't know if it's possible to send messages here so just add me in G+ (https://plus.google.com/+AleMoyano992)
Berak: what i need is to distinguish between several ant species.
Will Stewart: At first we want to use only one perspective (front) , then we will try to improve it so it can work with any perspective.