Edge AI

How to find, for HAR_IGN_WSDM network, accuracy.png, MatrixConfusion.png and Lossfunction.png. I can't find them in the utilities of AI_resources > models. Thank you very much Best regards Luigi Bibbo'

[STM32H745I-DISCO] Validation on target

Hi,I've some questions about validation on target, we meet some issues.Two ways are offer to do that, with or without automatic generate, compile, etc.1 - We tried the second method, and code generation stop due to not enough internal RAM. We are not...

X-CUBE-AI: error while analyzing .tflite model

I have followed this link https://wiki.st.com/stm32mcu/wiki/How_to_use_Teachable_Machine_to_create_an_image_classification_application_on_STM32 for image classification using teachable machine ,after model is created, in inspecting the model using ne...

X-CUBE-AI error while analyzing a Keras model in Ubuntu

The model I've been using for a year in CubeIDE 1.6.0 on a laptop cannot be analyzed now on a PC in the latest CubeIDE 1.6.1 while using the same X-Cube-AI v5.1.2. (I've tried the most recent v6.0.0 - the same error.)Ubuntu 20.04 (both laptop and PC)...

Problem implementing some layers of ONNX model although the documentation states that these layers are supported.

Although the documentation seems to support layers such as Unsqueeze and LogSoftMax for ONNX models, the X-Cube AI software gives errors such as1) INTERNAL ERROR - Unexpected layer type Unsqueeze2) TOOL ERROR - Unsupported layer type LogSoftmaxwhen a...

STM32H747-DISCO- STM32AI TOOL

Hi, Am trying to do image classification on STM32H747I- Disco. I got some ideas in St Wiki page. I have created the quantized model using some images. Now I need to work in Stm32ai shell. I really don't have any idea about how to work in stm32ai to...

Validation Model : Failed to build on Cube IDE

Hi community,I meet an issue when I try to build my project.I'm trying to implement a tflite ANN on STM32H745XI disco board, and validate him on target.My model is too heavy to embedded FLASH, so I configure for an external FLASH use thanks to QSPI....

Regarding the FP-AI-NANOEDG1 function pack for the STEVAL-STWINKT1B. How can I write the results of the anomaly detection to the SD card (for example, like the datalog results) ? Thanks a lot.

Neural network implementation on STM32 via prompt and STM32 CubeIDE "TOOL ERROR: not enough values to unpack" (expected 4 got 3)

Hi,i'm trying to analyze a neural network on STM32CubeIDE. With this tool in analyze mode i've got an error message. So i tried to do the same using command line and the message error is: "TOOL ERROR: not enough values to unpack" (expected 4 got 3)We...

Why X-CUBE-AI update From Local fails?

Hi, I wan't update X-CUBE-AI from 5.0.0 to 6.0.0 package in STM32CubeIDE 1.6.1 from local. So I've downloaded en.en.x-cube-ai-v6-0-0-windows.zip on my desktop. When I try to Load Froma Local the package "This file is either corrupted or not a recogni...