I’ve developed software that uses openCV/python and a cheap usb endoscope to calibrate a CNC machine by recognizing the position of black squares printed on a white 4-foot x 8-foot banner and developing an error matrix. If interested, look here:
I’m looking at the possibility of offloading the processing and saw the Pixy 2. The software I developed uses routines from pyimagesearch to locate the squares and determine where the center is located. If all I need to do is accurately determine the center of black squares on white background (likely only one square will be visible at a time), is the Pixy2 up to the task? The camera will be maybe 3 inches away from the calibration pattern and the squares are 0.5 to 0.6 inches in size (we are trying different patterns). Because this is used for calibrating a CNC machine, it’s important it be very accurate at determining the center position.