Reduce false positives in blob detection

I use an Open MV M7 to do blob detection on a production machine to see if the objects the machine work with appears at the end of a tube. It sometimes detects blobs when there are none. I think it needs a calibration of the blob thresholds every now and then. The device the M7 connects to can’t run the IDE, and even if it could, it would be hard to get the two to interact via the IU of the IDE.

Is there a way to have a computer calibrate the color-thresholds of the blob detection algorithm on a blank background without using the IDE?

Yeah, you need to devise a serial protocol and send the camera the new settings via that serial protocol over one of the I/O ports.

Okay. I didn’t manage to find any documentation that explains how I point at the right variables/settings and pass the data.

Do you have a doc that explains it or maybe even a code example showing how it is done? (we work in python on the external device that the M7 sends to as well)

We wrote a whole RPC library for this reason. However, people have experienced issues with it going to the PC. It’s only really rock solid over I2C/SPI/UART/CAN to another MCU.

Anyway, you can also just roll your own UART protocol to tell the camera what to do. This really isn’t hard. But, if it’s your first time doing COMMs you may find it very hard.