2025-04-23 10:08 AM
I currently have the STM32H7 DISC0 and plan on getting the STM32N6 development board. What I can't seem to figure out is if it is possible to train a custom dataset model on my own in tensorflow then download it onto the STM32 for inference? I was looking at platforms like YOLOv11 and Efficientdet. Also, are there any detailed videos or instructions I can look at? Please advise. Thank you.
Marcus
2025-04-24 1:22 AM
Hello @marcus69,
We are working on such video.
For now, I would suggest looking at the ST Model Zoo and ST Model Zoo Services:
The first repository contains model that we tested and that you can deploy as is or that you can use with the second repository to retrain, quantize and deploy example application for the STM32H747 DISCO and STM32N657 DK (for most usecases).
You can either use model from our repository or your own.
I let you look at the documentation I link below but basically, with the model zoo services, you select a usecase (image classification, object detection, audio event detection etc) you edit a file user_config.yaml to described what to do (training, with this model, this dataset, this parameters etc) and run a python script stm32ai_main.py.
I don't think we currently support YOLOv11 for now (with the st edge ai core, which is the tool that will convert python to C code).
How to set up model zoo: https://github.com/STMicroelectronics/stm32ai-modelzoo?tab=readme-ov-file#before-you-start
How to deploy on N6: https://github.com/STMicroelectronics/stm32ai-modelzoo-services/blob/main/object_detection/deployment/README_STM32N6.md
Object detection scripts: https://github.com/STMicroelectronics/stm32ai-modelzoo-services/tree/main/object_detection
Object detection models: https://github.com/STMicroelectronics/stm32ai-modelzoo/tree/main/object_detection
Have a good day,
Julian