Edge AI

Ask questions and find answers on how to deploy, debug, and optimize AI models on ST microcontrollers, microprocessors, and smart sensors.

cancel
Showing results for 
Search instead for 
Did you mean: 

Forum Posts

STWIN Trigger USR button via Firmware

Hi, im using the FP-SNS-DATALOG1 firmware for measuring sensor data and another custom firmware for other ML tasks. Furthermore, i am implementing a bootloader to switch between both softwares.  I want to measure automaticaly and in order to do so i ...

TomRett by Associate III
  • 2621 Views
  • 8 replies
  • 0 kudos

Generated models in NanoEdge AI Studio

Hi everyone, I am working on an extrapolation task as part of a research project at the university and I compare the different models, in this case random forest models. I assume that the model complexity of the model is related to the size of the kn...

Resolved! Problem with Deploy.py from Model Zoo

 Hi,I am currently working on my Final Year Project, which involves implementing a plant disease detection model on the STM32H747i-Disco board. My project is inspired by an image classification example from the model zoo.Back in October 2023, I succe...

Deploy.png chew_0-1706091318533.jpeg
chew by Associate III
  • 2100 Views
  • 5 replies
  • 0 kudos

Resolved! Reading .dat file

I am using the STEVAL-STWINKT1B evaluation board to gather microphone data with a 192kHz sampling rate. The data is saved in a .dat file which i want to read and process. However, when reading the IMP23ABSU_MIC.dat file i get completely different res...

TomRett_0-1707058815173.png TomRett_1-1707059110204.png
TomRett by Associate III
  • 2685 Views
  • 2 replies
  • 1 kudos

How to deploy a quantized model on an embedded platform

Hello everyone, I have a pytorch model (mymodel.pth) obtained by running the mixed-precision quantization of this algorithm: * https://github.com/eml-eda/q-ppg. I followed the instructions in the readme section.Python models are not natively supporte...

A self-trained model takes too much flash

Hi, I have a self-trained image classification model with two classes with dynamic range quantization(unquantized model is about 1.5mb and it is 500kb after quantization). And this model takes too much flash and ram after benchmarking. Is there any m...

undefined reference to `ai_emotion_model_create_and_init'

Hi, I am trying to create an STM32Cube.AI application using X-Cube-AI. I am following the documentation under Content 6. I trained my own model and configure the .ioc file as shown below. I wanted to get input from the ADC and feed directly into the ...

goldmine_0-1707289319048.png goldmine_1-1707289410395.png
goldmine by Associate
  • 1294 Views
  • 0 replies
  • 0 kudos

HardFault error in ai_platform_network_process()

Hello, I have implemented https://wiki.st.com/stm32mcu/wiki/AI:How_to_perform_motion_sensing_on_STM32L4_IoTnode this project. When I run the program I had a hard fault error when function enters  ai_platform_network_process().  I have incremented the...

Resolved! hsdatalog_to_wav.py conversion not working

Hi, I am currently using SensorTile.box Pro with FP-SNS-DATALOG2 FW. After acquisition, I would need to convert the data from the microphone from .dat to .wav and to do this I am using the hsdatalog_to_wav.py example provided. Unfortunately, the resu...

Gerva by Associate II
  • 2581 Views
  • 3 replies
  • 0 kudos