Troubleshooting ValueError: No subgraph in the model with PyCoral

Question

I've been working on a project using the PyCoral library to run inference on an Edge TPU. However, I've encountered a persistent issue where the make_interpreter function from the pycoral.utils.edgetpu module throws a ValueError: No subgraph in the model.

Here's a brief overview of my setup and what I've tried so far:

I'm on another Asus Tinker Edge T board after a year and trying to replicate the same project using the same model I have trained before.

I'm using a tensorflow>tensorflow-lite>edge converted .tflite model that was previously working.

The error occurs when calling interpreter.allocate_tensors() in my Python script.

I've confirmed that the model is compiled for the Edge TPU using the Edge TPU Compiler version 16.0.384591198.

I've attempted to install different versions of the pycoral library, but it seems only version 0.2.0 on PyPI and 2.0.0 is available througout using (pip3 install --extra-index-url https://google-coral.github.io/py-repo/ pycoral).

I've also checked system-wide installations and ensured that pycoral is installed in my Python environment. The model and related code have not changed, so it seems to be an issue with the environment or library versions.

The unfortunate part is that the card is toasted and there is no way for me to check the environment that it works well with.

Here's the relevant part of the code that's causing the error:

[python]
from pycoral.utils.edgetpu import make_interpreter

interpreter = make_interpreter('model_edgetpu.tflite')
interpreter.allocate_tensors()

I've considered the possibility that the TensorFlow version might be causing compatibility issues and in need of building from scratch, but I haven't been able to pinpoint the exact cause.

I've also thought about recompiling the model with the latest version of the Edge TPU Compiler, but I only have the final .tflite file.

I'm reaching out to see if anyone has faced a similar issue or could provide insights into what might be going wrong. Any suggestions on how to resolve this error or further steps I could take to debug this issue would be greatly appreciated.

Thank you in advance for your help!