I’m running into the following issue here:
I have a FastAPI server with a call connected to OpenCV and EasyOCR to read an image text, however the executions crashes whenever EasyOCR reader is called (which internally, uses a torch model). The last statement that I receive would be:
[Could not initialize NNPACK! Reason: Unsupported hardware]
I’ve read that this message is nothing more than a warning but still, it is the last message that I receive from the server machine (again, a docker container within vmware without GPU).
Can someone provide some light?
I’m using torch cpu-only libraries and I have also tried to downgrade Python to 3.8, which is currently my current version.