The complexity of the image processing problem may depend how much variation there is in the images (registration, light levels, etc) but if you know what you're doing, OpenCV in Python is a convenient framework to use. (You will need to combine it with Picamera2, rather than try to grab video directly in OpenCV. Both packages are freely available and well documented.)
I'm a bit old-school, but I think I would not use a neural network especially if you need quantitative results (e.g. relative intensities of the bands) as well as a classification result.
Without knowing more about the application, I'd suggest a Raspberry Pi 4 or 5 with at least 2GB or preferably >=4GB of RAM (smaller Raspberry Pis can struggle with imaging when combined with a GUI), and a HQ camera (12Mpix) with a lens appropriate to the size and distance.
BTW if this topic disappears it'll be because it's moved to "Camera board".
I'm a bit old-school, but I think I would not use a neural network especially if you need quantitative results (e.g. relative intensities of the bands) as well as a classification result.
Without knowing more about the application, I'd suggest a Raspberry Pi 4 or 5 with at least 2GB or preferably >=4GB of RAM (smaller Raspberry Pis can struggle with imaging when combined with a GUI), and a HQ camera (12Mpix) with a lens appropriate to the size and distance.
BTW if this topic disappears it'll be because it's moved to "Camera board".
Statistics: Posted by njh — Thu Jan 08, 2026 10:57 am