Due to high demand expect some shipping delays at this time, orders may not ship for 1-2 business days.
0

Tensorflow Lite on EdgeBadge with Conv2D files
Moderators: adafruit_support_bill, adafruit

Please be positive and constructive with your questions and comments.

Tensorflow Lite on EdgeBadge with Conv2D files

by ankixyz on Fri Oct 22, 2021 12:06 pm

Dear everyone,

we are struggling with deploying deep neural networks to the EdgeBadge that contain Conv2D layers. We are relying on the sine model sample script ("hello world arcada") as used in the book TinyML. When we are using Dense layers only, everything works just fine. However, when we introduce ordinary Conv2D layer in the model and keeping the whole deploying pipeline constant, the interpreter cannot load the model file from disk.
We also tried to directly encode the TensorFlow lite model as C-array. Again, this also works for Dense layer based models, but not for ones with >= 1 Conv2D layer.

Is there something we are missing or is the support incomplete?

Thank you so much for your help.

Best,
anki

ankixyz
 
Posts: 1
Joined: Fri Oct 22, 2021 12:02 pm

Please be positive and constructive with your questions and comments.