I need the update rpc library for the rt 1062 for a project of mine, I saw on the github that it’s been updated just not released, where in the file system of an installed openmv ide, can I find the source so I can cut + replace from the github?
Edit: Actually I saw that it was in the development release on github but only for the rpc_spi_master, but it doesnt seem to be done for all of rpc (speciffily rpc_usb_vcp_slave()) , quite yet, since it still has the “no module named pyb”.
still would maybe good to know for future reference
Yeah, because of the machine module limitations at the moment there are no more interfaces we can add support for except CAN.
If you want to modify the RPC library code just copy the past the file here on your camera and make the changes: openmv/scripts/libraries/rpc.py at master · openmv/openmv (github.com)
Name it something different, though, as the built-in one will always be loaded first if it shares the same name.
If you want to send data using the OpenMV Cam to the PC you can just do print() in your script. As for receiving data, in non-debug mode, the system will present a REPL interface like every other OpenMV Cam board.
But, yes, due to the machine module lacking the VCP uart feature we do not have the VCP uart on the RT1062 at this time.
I was running different models on the board but ultimately found that at a decent amount of accuracy, 62%, I was only getting about 2.5fps with that board and the global shudder cam, so I wanted to send the images to the computer to run It on way beefier hardware. Basically to use it as a glorified webcam. I figured it wasn’t as easy as replacing machine with pyb as it sometimes is based on the lack of online information. My only other option is asking way I would get such low fps on the board, but I think that relies more on the size of the image I originally chose (QVGA).
Use the pyboard.py script to control the board and pull data from it via a PC: micropython/tools/pyboard.py at master · micropython/micropython (github.com)
Or our pyopenmv.py tool: openmv/tools/pyopenmv.py at master · openmv/openmv (github.com)
Which uses the debug protocol so you can start/stop the board, send scripts, and pull the frame buffer.
The speed at which the frame buffer is pulled will be incredibly fast on the RT1062. So, you can have a script stream video. When you need to change settings just stop the script, and send a new script with different settings.
Can’t comment on why QVGA is slow. That should be at 80 FPS with the global shutter camera. If you are running a huge model on it though that’s just how long the processor takes to run the model.