Edge AI

..

I am using the STM32H745ZI-Q Dual Core MCU. My workflow is to generate the AI library using a Single Core processor and then copy the project into my Dual Core. Now I am trying to use External Ram Configuration (STMCube) to map the activations NOT into a external RAM but into another Bank (AXI SRAM for example. Address: 0x30000000) but without any luck. Is there any way I can allocate the Activation buffer on another RAM Bank?

Hi, everyone have someone know where can find the voice recognition of STM32L1 application?I just need few Japanese vocabulary to identify command, but I didn't find it. AWS didn't match my request. Could you please help me?

How do I run larger ML models (greater than 10MB)? Is interfacing external ROM with QSPI and memory mapping it alone sufficient?Can I use an SD card instead? Can the memory card be memory mapped to internal flash? Can I paste the weights on the memory card using my PC and have the ML model request the SDMMC everytime it needs to access some data?

Hello,I am undergraduated student in South Korea. Right now, I've ran into a little problem during my research. So I'm uploading this message becasue I believe that you guys can help sloving this probelm. I wanted my model to work in right way while using sensor tile but even though I've tried a lot of time, it didn't work properly. Simply speaking, I quantized my model with npz file, which needed verification. I converted quantized mdeol file through code generator (STM32 AI5.0) and changed the converted result (C,H file) with function pack file (fp-ai-sensing1). At last, code about part which needed softmax didn't work due to absense of output meta info. So like a blue part in attached photo, I arbitrarily did 'hard coding.'I tired asc on 2 model (1.model provided by function pack 2. model I've exercised) and it showed conflicting results. In case of model provided, it worked properly when went through same progress. However, in case of my model, even though I checked that it wo

When using the CubeMX code generator, I am not able to include the Cube AI library when selecting STM32H745ZI-Q. Is there a way to use one of the H7 Family Dual Core MCU's with the CubeAI library? Or is it possible to generate the necessary library using a single M7 or M4 cortex MCU and then copying the generated files into the Dual Core project?

Hello all,I have bought a SensorTile.box which has LSM6DSOX, an IMU with Machine Learning Core. I am seeing some decision tree-based examples/implementation of LSM6DSOX.If I train a new model for my application, My query is how to load that model into LSM6DSOX of SensorTile.box. General examples of LSM6DSOX says I need to insert the LSM6DSOX (DIL 24 socket) in an ST MEMS adapters (STEVAL-MKI109V3) and use Unico tool. How to do that for SensorTile.box?Another query is whether any other low-cost option available to load a new decision tree model into the LSM6DSOX chip without buying that STEVAL-MKI109V3.Thanks in advance.

Hello, I am under graduate student in Korea and I have a question to solve a little of problem.I tried so many time to classify at sensor tile.In my case, I quantized my model with validation npz and converted to C code by codegenerator which is as known as STM32_AI 5.0.0. After converting, I changed asc.c asc_data.c and .h file too at fp-ai-sensing1 source. Finally I added the softmax code because code doesn't exist output meta data in this function pack.​I tried asc(acoustic sound classification) about two models(provided model by function pack and made new model by me) and I get the results. First result is perfect. It works very correctly. Second result displays same value for some classes.​I think cause of this problem is feature scale vector is wrong. This made for proviede model. So I want to make new feature scale vector for my model.Would you please help me to solve this problem?​Thank you for read this question!

ing the sensor box. But as soon as the intensity value changes the graph and data collection freeze. In fact, I noticed before I finished the example and was just plotting the three accel values that the graph would freeze occasionally, but never reproducibly. Is this a bad USB cable, MCU crashing, LSM6DSO)X crashing? Any idea why I can;t get the simple webinar example to work robustly?

Hello!My problem is that I don't see the components in the Additional Software menu.I did everything strictly according to the instructions specified in the official video "Getting Started with STM32Cube.AI". Components installed as shown.CubeMX reinstalled, the computer rebooted, but it did not help. Touching Refresh button also has no any effect. What could be the problem?

..

..

..

..

..

..

..

..

..

..

..

..

..

On May 12th an interesting Webinar was presented on the smart use of the Machine Learning Core embedded in the X-family IMU sensors, such as the LSM6DSOX mounted on the SensorTile.box board.The main example proposed during the webinar is related to the use of the SensorTile.box together with ST tools (Unico, Unicleo and Algobuilder softwares) that can be freely downloaded from the st.com website.The video examples available in the webinar documentation show step-by-step the whole process for the implementation of the artificial intelligence function on the SensorTile.box tool, from the data acquisition and the data-set creation (a "Mode 1" use of the SensorTile.box board) to the direct configuration of the LSM6DSOX sensor and data acquisition through Unicleo and Algobuilder softwares (a light but very effective example of "Mode 3").This can be useful to have a practical view on the use of the Mode 2 / Mode 3 of the SensorTile.box for th

Hi everyone,I am trying to implement an image recognition application on an STM32L475. I succeded implementing a Keras model with Cube AI and wanted to reduce the memory footprint with a quantization model. I used the following code to quantize the keras model according to a question I found on this forum.# Model_s is a trained model in the same python script converter = tf.lite.TFLiteConverter.from_keras_model(model_s) converter.target_ops = [tf.lite.OpsSet.TFLITE_BUILTINS_INT8] converter.inference_input_type = tf.int8 converter.inference_output_type = tf.int8 quant_model = converter.convert() # Saving the model open("Models/converted_model.tflite", "wb").write(quant_model)When I try to analyze the model Cube AI reports the following error:Analyzing model Neural Network Tools for STM32 v1.2.0 (AI tools v5.0.0) -- Importing model INTERNAL ERROR:  Creating report file C:\Users\<user_name>\STM32Cu