Edge AI

Hi all,​I tried to porting AI model which I made with cubeAI but it seems like not work.​I succeded [validation on target] once but I don't know how I could do this.​​​​When I run [validation on target], it respone that it cannot find NetworkRuntime720_CM4_GCC.a.​ However NetworkRuntime720_CM4_GCC.a is already exist and even I add path of that library. ​Is there someone why it cannot work?​EnvironmentNucleo-L432STM32CubeIDE 1.9.0Cube-AI 7.2.0Window10 Edu 

Hi is it normal this error when IA Analyzer is used?Any error appears... thanks for your time.

Noob question - can I use AI with the STM32WL SoC ?I want to train a model and then use STM32Cube.AI to put it on STM32WLE5CC. This is so I can use AI on the end device and transmit data using LoRa all using the one chip.Is this possible ?Many thanks in advance.

Hi,I'm working on an application on a STM32WB55 which is sampling sensor data with 52Hz and feed it into a Keras model. Both the raw sampling sensor data as the output of the model will be saved to external flash memory (every 64 samples). A 52Hz interrupt DRDY signal will trigger to start a sample. The device will sample the data for several hours a day. At the end of the day all data will be offloaded over BLE. So in the project the BLE stack is included as well. Now I wonder if I can add some low-power features to this project. I would like to have some advice:My clock (HSE/ sysclock) are now running on 32 MHz. Is this something I can lower when running an AI model?Any advice on which mode is possible (like low-power RUN/SLEEP/STOP) ?Is it possible to shutdown CPU2 (BLE core) while CPU1 is running the AI code / sampling data?I hope one of you can help me with one or more of above questions.Thanks already!

I realize you can use the stm32ai tool to find the SRAM/Flash usage of the model. However, this does not include the "Library" size. This "Library" size is only calculated when you click analyze in Cube IDE. See above image. How do I calculate "Library"?TLDR: I need to calculate total SRAM/Flash usage of a TFLite model, INCLUDING Library memory usage.

..

Hi friends, I am now trying to bring .h5 model to my board. So just following tutorial from st.​(tutorial given by stm 32 cube ai..) cannot add links..ㅠㅠ​But, I have problem following these tutorials.​​Really don't know why my program doesn't show additional softwares as tutorial....​(Changed program's versions & Changed board versions already...)​​But they didn't work at all.... plz help me..!!

Hello.I evaluate the SensorTile Kit with the following combinations.Core system STLCS01V1Cradle expansion STLCX01V1​When I connect to the Android app "ST BLE Sensor" and select "AI Data Log" to collect learning data, I get an error saying "SD Card is missing".What does this mean? How can I resolve this?Regards, Gotoda

..

..

Dear ST Engineer, Hello, I am doing sound classification (AER-CNN-KERAS) artificial intelligence, hardware STM32H743IIT6 (400M) external extension SDRAM (W9825G6KH-6) This is a 16-bit data width 32M bytes memory rate 100M After downloading the model to STM32H743, it is estimated that a speech (a piece of audio) recognition will take about 40 seconds, which is too slow to accept. I would like to ask what is the reason for such a long operation time?  My analysis suggests that external SDRAM is the main reason, the model requires 9.31MiB of space all outside. But it should not be so slow. Is there any way to speed up the operation? Is it that STM32CUBEMX_AI has not recognized that SDRAM is a 16-bit memory? And use it as an 8-bit? If I extend 32-bit SDRAM, can I speed up the computation?

I want to know if I can use different features for different decision trees. I am using an LSM6DSOX. Since I can configure up to 8 decision trees, can I have each tree use different features, such as tree#1 using mean and variance, while tree#2 using the XYZ values of the accelerometer instead.Any help is appreciated. Thanks!

I am trying to go through the Quick Start Guide and am able to successfully load the FP-AI-PREDMNT2.bin firmware on my STEVAL-STWINKT1B and have the ST BLE sensor app on my iPhone (version 4.14.6). I am able to connect to my board (although my board says PM2_110 instead of PM2_100). But when I open Certificate Request I don’t have an option to select "Predictive Maintenance" and have a different screen that says "Done Certificate Management" and shows the device ID of my STEVAL-STWINKT1B. I don’t get to the other screens and can’t sign in with my myST credentials and do the rest of the setup. I get an error message on the ST BLE Sensor app "The operation couldn't be completed. (AssetTrackingCloudDashboard.App- AuthError error.1)'

I built the LSTM network with keras 2.6.0 and generated the MDK project with CubeAI 7.10. However, in running the program AIInit() reported an error with an error type of 33.Do you know what the problem is? Thank you very much!

I built the LSTM model using keras 2.3.1 and generated the MDK project with CubeAI 6.0. But the LSTM model cannot be validated and parsed. Do you have any examples of using LSTM? Thank you!

Hello all,Is it possible to create, inject data and perform the inference with just my STM32F4-DISCO without any connection with my PC (of course I must beforhand flash my model into my board). And how would you do it, I tried to integrate the code from "How to run locally a c-model" in my file but it doesn't seems to work. I thanks you in advance !The code :int ai_network_***(void) { ai_u8 activations[AI_NETWORK_DATA_ACTIVATIONS_SIZE]; ai_u8 in_data[AI_NETWORK_IN_1_SIZE_BYTES]; ai_u8 out_data[AI_NETWORK_OUT_1_SIZE_BYTES];   /* AI buffer IO handlers */ ai_buffer *ai_input; ai_buffer *ai_output; ai_handle network = AI_HANDLE_NULL; ai_error err; ai_network_report report;   /** @brief Initialize network */ const ai_handle acts[] = { activations }; err = ai_network_create_and_init(&network, acts, NULL); if (err.type != AI_ERROR_NONE) { printf("ai init_and_create error\n"); return -1; }   /** @brief {optional} for debug/log purpose */ if (ai_network_get_

Hi, I created my own model and tried to import it into the STM32L476 MCU.When I execute the ai_network_run( ) function, I got the hard fault interrupt.Here is the function below. ( I refered the code from here :https://wiki.st.com/stm32mcu/wiki/AI:How_to_perform_motion_sensing_on_STM32L4_IoTnode)static void AI_Run(float *pIn, float *pOut) { ai_i32 batch; ai_error err;   /* Update IO handlers with the data payload */ ai_input[0].data = AI_HANDLE_PTR(pIn); ai_output[0].data = AI_HANDLE_PTR(pOut);   batch = ai_network_run(network, ai_input, ai_output);   if (batch != 1) { err = ai_network_get_error(network); printf("AI ai_network_run error - type=%d code=%d\r\n", err.type, err.code); Error_Handler(); } }I have checked the 'ai_input' and 'ai_output' values and they all pointed well. Here is the main loop.while (1) { AccelReadValues(0x32, 6);   x = ((AccelData[1] << 8) | AccelData[0]); y = ((AccelData[3] << 8) | AccelData[2]); z = ((Accel

Hello Community,I am trying to integrate a deep neural network into the micro-controller with the help of STM32 Cube IDE. When I try to analyze the model I get the following errors attached as a screenshot. I have to enable the special operations while converting the model to the .tflite model. How can I integrate the model with special operations enabled and analyze it without errors?Best Wishes,Siddhesh

Hello,I'm trying to send data and retrieve the output from my stm32f4-DISCO. I'm using a python based code and specifically stm_ai_runner. The code is rather simpler but I can't access the board because even runner.summary() doesn't work. My board is flashed with the network :And this is what my code looks like :And the terminal : I thank you in advance,Numa

I suspect it's an issue with the stm32ai binary not being compliant with Mac M1, but perhaps someone can guide me towards the solution.I tried the simplest X-CUBE-AI examples, such as the one with the sine wave and the one with HAR. Even when I take networks pretrained by others, loading them in the STM32CUBE gives me "Analyze fail on the AI model" every time. I tried running stm32ai directly from the command line and I got "illegal hardware instruction" error. Anyone running X-CUBE-AI on Mac M1, could you please help?Thanks!

..

I am using the model from the posted page (model.5h). It's a Keras model. When i try to analyze it i get error: Analysis status: error.Validation metrics: No data to display, please perform a validation first.It happens with a single model which i try to analyze. How to fix it? I use stmcubeide 1.6.1 , and X-CUBE-AI 7.0.0 package, win 10. I tried also use options validate on desktop and target but that's not help. Also tried with old version of X-CUBE-AI (5.2.0). I put some screens with an error.Could it be a problem with the model itself or with the environment?

I tried Cube AI like youtube https://youtu.be/grgNXdkmzzQ. But I got some difficulties.I use Cube MX 5.0.1,Cube AI 3.3.0.1.   Invalid Network Error In mcu selection, I selected Github Model file and pushed analyze button in the funtion item. Then I got Invalid network error and difficult to analyze the model.In the youtube, here : https://youtu.be/grgNXdkmzzQ?t=2172.   Unable check AI in the Pinout & Configuration Setting I selected F7 MCU. But the item of Artificial intelligence doesn’t appear in the Application software. I tried to display the item by checking AI in the additional software. But the item doesn’t appear in the Application.In the youtube, here : https://youtu.be/grgNXdkmzzQ?t=316Anyone has solutions?Thank you.

Hello,I created my model using Colab and saved it as .tflite (tensorflow version is 2.8.2)I want to use STM32H743AII.However, when I analyze it using XCube.AI I got the error below: [AI:network] Neural Network Tools for STM32AI v1.6.0 (STM.ai v7.1.0-RC3) [AI:network] INTERNAL ERROR: 'FlexTensorListReserve'I tried to downgrade TensorFlow 2.6 e Keras 2.6 but I got same error.Annexed the .tflite model for your reference.Can you pls help me to understand the reason of this error?can I fix it from my side?Thanks for your support

Hi,I'm trying to import a .h5 model generated by an online AI platform in X-CUBE-AI.7.1.0, receiving the error:E010(InvalidModelError): Model saved with Keras 2.7.0 but <= 2.6.0 is supportedWill there be any support on this version? Is there any workaround to e.g. convert a .h5 to an older version?