How to Use a .tflite File with MicroPython for CNN Face Recognition?

[A0GV]

I am working on a project that involves using MicroPython and a Convolutional Neural Network (CNN) for face recognition. I already have the model saved in a .tflite format. However, I am struggling to find a way to import and use the .tflite file in MicroPython.

I have tried:

Importing TensorFlow (import tensorflow as tf), but TensorFlow isn't supported in MicroPython.
Searching online for solutions or libraries that might work with MicroPython, but I haven't found a clear answer.
Are there any alternatives or lightweight libraries available for MicroPython that can run a .tflite model? If not, are there other approaches I can consider for deploying face recognition in MicroPython?

Any advice or suggestions would be greatly appreciated!

Also, I have the possibilte of change the .tflite to an .h5

Imi using an ESP32 PSRAM Timer Camera X from M5

[jonnor]

OpenMV has a Tensorflow module. To use that one has to use the OpenMV "distribution" of MicroPython (not the official downloads). See their documentation

Otherwise, there is a CNN module in emlearn-micropython, which is distributed as a drop in .mpy module. I have tested it a bit on ESP32. This is based on TinyMaix, which takes a Keras/Tensorflow network or a .tflite file, and converts it to their format. See the examples at https://github.com/emlearn/emlearn-micropython/tree/master/examples/mnist_cnn

[A0GV]

Thanks for the guide. I will update with my progress.