Automatic reload of a trained model

Discussion related to "under the hood" OpenMV topics.
Naig
Posts: 2
Joined: Wed May 20, 2020 9:54 am

Automatic reload of a trained model

Postby Naig » Wed May 20, 2020 10:15 am

Hi all,
I'm currently working with an OpenMV H7 on neural network. The network is dynamically re-trained on a main board so I have to automatically upload new model on H7 every time it is available on main board.
I guess I have to modify the firmware to do this so I would like to ask you some hints.

Thanks for all replies
User avatar
kwagyeman
Posts: 4091
Joined: Sun May 24, 2015 2:10 pm

Re: Automatic reload of a trained model

Postby kwagyeman » Wed May 20, 2020 11:06 pm

With the TensofFlow module you just delete the model object from RAM and accept a new model.

How is the model getting onto the OpenMV Cam?
Nyamekye,
Naig
Posts: 2
Joined: Wed May 20, 2020 9:54 am

Re: Automatic reload of a trained model

Postby Naig » Thu May 21, 2020 5:07 am

The model is a quantized model output from STM32CubeAi.

Can you explain me better?

Thank you!
User avatar
kwagyeman
Posts: 4091
Joined: Sun May 24, 2015 2:10 pm

Re: Automatic reload of a trained model

Postby kwagyeman » Thu May 21, 2020 7:55 pm

You need to edit the firmware yourself.
Nyamekye,

Return to “Technical Discussion”

Who is online

Users browsing this forum: No registered users and 4 guests