We have created 5 DL models using tensorflow and converted to .tflite. Then using tflite-micro specified the architecture required i.e Texas Instruments CC1352P7-1 LP using ARM cortex M4F for deploying to hardware. We compiled the code and flashed the binary file to hardware using the Texas Instruments IDE Code Compressor Studio.
However, while trying to run the code we are getting a faultISR() error - the program is going into an infinite loop as specified in the code when facing an interrupt issue. However, why would this occur ? If we are using the tflite-micro library and specified architecture then why would we be running into memory issues on the hardware?
How do we identify what is causing the error and resolve it?
Also tried the helloworld file itself. Compiling correctly but on execution - same problem - faultISR(). Unable to understand this and if it is in the example file itself I would assume it is a bug in the original code itself?
My team does not have any experience with deploying AI models to hardware and/or C++ programming. We are purely Data Scientists who use python and associated libraries for our use cases. This is the first time that we are using tflite-micro as that is our clients use case.
Any help and clues will be appreciated.