Usnig Caffe to detect objects (not specific, just "objects")

[A Madey]

Hi there,

I'm new to Machine Learning, but I have been given a task to detect a certain set of objects (in this case, insects), within a particular spatial range, by use of a camera, hooked up to a computer running a Machine Learning model.

The insects are small, and will not be distinguishable from leaves or other objects in the air (since we will be testing outside), and further measurements (not by camera, or ML) will have to determine what proportion are insects, what proportion leaves, what proportion birds, etc.

The question I have is: does it even make sense to pursue a Machine Learning application, in which I am looking to tabulate objects, without attempting to classify them, and only to count how many there are? If so, would Caffe be the right option, or would it be better to use a different tool?

Thanks! Sorry for the long question; just trying to figure things out.

Best,

Andrew