Edge AI

[Ritesh1]

How to check the classification accuracy of image classification models on STM32H7 ?

 

 

 

 

Ritesh

 

 

 

 

 

 

 

 

 

 

 

[Julian E.]

Hello @Ritesh1 ,

You can use the ST Model Zoo using the benchmarking operation mode. Please find the documentation about it here: stm32ai-modelzoo/image_classification/src/benchmarking/README.md at main · STMicroelectronics/stm32ai-modelzoo · GitHub

 

First quantize your model and then do a benchmark on your test dataset.

You can train, quantize, evaluate, benchmark and deploy a model with the model zoo. 

Everything is clearly explained in the github repository.

 

If you already have deployed your model on your target, you can create buffers directly with your image that you put through your model, or develop other way to pass the images to the model (sd card for example).

 

You can also take a look at the ST Edge AI Dev Cloud. it is very easy to use.

Please find the documentation here: https://wiki.st.com/stm32mcu/wiki/AI:Getting_started_with_STM32Cube.AI_Developer_Cloud 

 

Have a good day,

Julian