Edge AI

Hello, I'd like to know if ST could provide software package or any tutorial to realise the Vibro-Acoustic Sensor Fusion demonstrated in this video https://www.youtube.com/watch?v=bi4NenlnR-k? I have an STEVAL-STWINKT1B on hand. Thanks in advance.  

In ST Edge AI Developer Cloud I generated the C library of a ST_LSM6DSO16IS_ispu_wand_dataset with platform selected as ISPU. I got many library files in which the major APIs are in the network.h and its c file. I need an API manual or documentation to apply this in tho my program or an example of implementation is enough.

Hello, I generated the C-code for my AI model using Cube AI, and I'm struggling with how to correctly set up the function calls and everything to initialize & run my model on Cube IDE. I made these first lines to verify if the returns were working fine for the initialization part: ai_handle myNetwork = NULL; ai_network_params params; ai_error err; err = ai_network_create(&myNetwork, AI_NETWORK_DATA_CONFIG); if (err.type == AI_ERROR_NONE) printf("Network instance created\r\n"); if (ai_network_data_params_get(&params)) printf("Valid configuration is present\r\n"); if(ai_network_init(myNetwork, &params)) printf("Network initialized\r\n"); When debugging, I could get the proper return for the ai_network_create and ai_network_data_params_get, but I wasn't being able to initialize the network. Going further, I saw inside ai_network_init declaration: ai_bool ai_network_init( ai_handle network, const ai_network_params* params) { ai_network* net_ctx = ai_platform_network_init(n

Hello , I have a problem when trying image classification with STM32H743ZI2+ B Cam Omv,My generated code from ST_Model _Zoo build with Camera DCMI Mode + USB Display  is oke, but afer i flash, My display stop here , seems like it can't get new frame and stay at function while infinitie becausce new frame ==0,I want to ask how to capture image and make it run like exampleBest Regards!      

I am using the STEVAL-STWINBX1 development board with the FP-SNS-DATALOG2 software.I am struggling to configure PB1 as a digital input for an external interrupt (which is on pin 12 of the expansion connector CN1)... I start by opening the .ioc file in STM32CubeMX...\STM32CubeFunctionPack_DATALOG2_V2.1.1\Projects\STM32U585AI-STWIN.box\Applications\DATALOG2\mx\STEVAL-STWINBX1_FP-SNS-DATALOG2.ioc Then I change PB1 to be an external interrupt, with a pull up resistor and falling edge trigger...I then generate code in STM32CubeMX and refresh the project in STM32CubeIDE.Am I correct in saying the interrupts are in the file stm32u5xx_it.c (as shown below) ?Confusingly, there are two stm32u5xx_it.c files in the project in the following folders...STM32CubeFunctionPack_DATALOG2_V2.1.1\Projects\STM32U585AI-STWIN.box\Applications\DATALOG2\mx\SrcSTM32CubeFunctionPack_DATALOG2_V2.1.1\Projects\STM32U585AI-STWIN.box\Applications\DATALOG2\Core\SrcThe file in the DATALOG2\mx\Src

Hello everyone, I am working on a project using nanoedge studio ia in rating N, I want to check my project with cube ide, I can make a view with tera term or other tool, which shows the result corresponding to my database. I used stm32L475vgt6; The problem is that there is also no display with a simple code that says printf(hello World)

I have built and trained a recursive model with my data and generated the `.h5` and `.tflite` files. While I have been able to optimize the `.h5` file successfully, I am facing difficulties quantizing it. Below is the error I receive when attempting quantization:   ////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////////Executing with:  {'model': '/tmp/quantization-service/24d7cfa5-3e25-4be7-b847-3fc7bf2b6d81/RNN_16_model.h5', 'data': None, 'input_type': tf.float32, 'output_type': tf.float32, 'optimization': <Optimize.DEFAULT: 'DEFAULT'>, 'output': '/tmp/quantization-service/24d7cfa5-3e25-4be7-b847-3fc7bf2b6d81', 'disable_per_channel': False} No data specified, enabling fake quantizationConverting original model to TFLite.../app/quantizer/cli.py:53:1: error: 'tf.TensorListReserve' op requires element_shape to be static during TF Lite transformati

I am using the STEVAL-STWINBX1 development board and have downloaded the FP-SNS-DATALOG2 software. I am trying to get the WiFi working using the ST BLE Sensor apps, but the option 'Set Wi-Fi credentials' is greyed out.     I have even tried the classic app...   I have also tried it with the default firmware BLEDfFw, but again the 'Set Wi-Fi credentials' is greyed out. The user guide shows to click on the 3 dots in the right corner of the app....   But in my app, there are no 3 dots in the top right corner... How do I get the WiFi connection working with this board ?

I am using the STEVAL-STWINBX1 development board and have downloaded the FP-SNS-DATALOG2 software to the board. Is it possible to add another channel, in this case a digital input that will be logged to the SD card ? I need to synchronize the data logs from different DAQ systems including the STEVAL-STWINBX1 I was thinking if the STEVAL-STWINBX1 could log a GPS PPS (Pulse Per Second) digital signal, then I could manually (offline) sync its log with the logs from the other DAQ systems. Or is there some other technique I can use to synchronize all these logs from different DAQ systems (e.g. IRIG, NTP, EtherCAT, PTP).        

I am using the STEVAL-STWINBX1 development board and have downloaded the FP-SNS-DATALOG2When I open the help file FP-SNS-DATALOG2_Package.chm (in STM32CubeFunctionPack_DATALOG2_V2.1.1\Documentation) it appears to be empty other than the index...Why is the help file empty, how do I get it to work ?

Is there a MATLAB script to plot and export the .dat files saved by the STWINBX1 datalogger ?

Not be able to compile as it asks for ai_platform.h? But the file exists: Middlewares/ST/AI/AI/include/ai_platform.hCopied into teh Inc folder but then I got this bps_ai.h is not found.I am using NUCLEO_H743ZI on Mac.Any suggestions?Thank you in advance.

i downloaded stmzoo file from github. followed the steps in how to install stm modelzoo. created a virtual machine envirement. installed all the requirements (manually because i kept having errors when installing them all automatically using the command install -r requirement.txt) . however . when i want to deploy it using the command python deploy.py after changing my place to the deployement folder . i keep get this error that say no module found with the name common-deploy although it exist in the modelzoo folder under common folder . why??(windows)

I'm currently in the process of choosing an IMU/sensor hub for activity recognition, and am stuck between the ISM330IS(N) and the ISM330DHCX. I want to use the sensor hub function and link external sensors, and run (H)AR algorithms on the IMU itself to offload my main MCU. My question: I don't fully understand the difference between the DHCX's MLC and the IS(N)'s ISPU. Which one is more powerful? I definitely get more freedom using the ISPU, but the DHCX's datasheet does not provide any performance/memory data for the MLC. Am I right in the assumption that I could emulate any algorithmic processing the MLC can do using the ISPU, with the added benefit of dynamic reprogramming and custom code? I found this article (https://community.st.com/t5/mems-and-sensors/which-sensor-embedded-feature-should-i-choose-for-my-application/ta-p/49671), which implies the ISPU is more powerful than the MLC, but may have more power draw. Is this correct? Best regards,FG

Hello Developers,  We are currently developing a product for industrial machines condition monitoring and predictive maintenance application. For evaluation purpose we have bought "STEVAL-STWINKT1B" for the same. We are using the "en.stsw-stwincell" cellular example for publishing stwin's data to the aws cloud. In this example for vibration analysis, on board IIS3DWB sensor is used with default sensor parameters (i.e stated below) and the Time Domain Analysis output (i.e Acc RMS, Acc Peak and Speed Rms) in all 3-axis is published on the AWS cloud after the Time Domain and Frequency Domain analysis has finished i.e after TACQ (5 seconds). And for reference measurements we are using SPM Portable Vibration meter (VIB-15) based on analog sensor, generally used in industries for measurement of  true acceleration and velocity values in 1-axis of machine for condition monitioring.    Parameters used in Cellular Example: IIS3DWB Configuration Acc ODR:&nb

这是我想导入的数据集，希望模型能起到这样的效果：文档中是各个时间点的温度值，我需要模型能学习一天中的温度变化范围，当温度偏离较大时检测为异常，请问我在导入数据时，为何显示无法import？（Import按钮是灰色的，无法点击）

sampling finder (SF) for imported signal file is greyed out, please advise

I compiled a 1C library using nanoedge ai studio and verified the sample successfully using the emulator. I also successfully added it to my Keil project, but the result was always 0 when I ran it. My code: Error_code = neai_oneclass(input_user_buffer, &oneclass_result); onecclasss_result is always 0, regardless of the data. Why?

I am using the STEVAL-STWINBX1 development board and have downloaded the FP-SNS-DATALOG2 software to the board. When I use the board in stand alone mode, the data is saved to the SD card and the folders are sequentially named as DL2_00001, DL2_00002, DL2_00003 etc. When the board is connected via USB to a computer, the data is streamed and saved to to the computer and the folders are named with the year/month/date/time (YYYYMMDD_HH_MM_SS). Is there anyway to name the folders with the year/month/date/time (YYYYMMDD_HH_MM_SS) when using the board in stand alone mode and logging to the SD card ?  Whether that means implementing a real time clock or receiving a UNIX time from some external source ? I need to synchronize the data logs from different DAQ systems including the STEVAL-STWINBX1, so a folder named with year/month/date/time is a requirement. Note: This also relates to my post about synchronizing different DAQ systems... https://commun

..

Hi,I've joined introductory imbedded AI course on Udemy, unfortunately it is more than 4 years old using AI pack version 3.4.0 and I cannot get support from the instructor.Please find attached a presentation with screen shots of all steps used to generate the project and compiler messages after each corrective action until I can no longer do anything and stuck.Thank you in advance   

Hello everyoneI hope someone can help meI have an ML model that consists of 2 inputs as shown in the picture. Is the code for the inputs correct?   float ML_input[WINDOW_SIZE*NUM_POINTS]; float Diff_input[WINDOW_SIZE*NUM_POINTS - NUM_POINTS]; static ai_float activations[AI_NETWORK_ADEM_DATA_ACTIVATIONS_SIZE]; static ai_float in_data[AI_NETWORK_ADEM_IN_1_SIZE_BYTES]; static ai_float out_data[AI_NETWORK_ADEM_OUT_1_SIZE_BYTES]; ai_handle adem_network = AI_HANDLE_NULL; static ai_buffer *ai_input; static ai_buffer *ai_output; ai_i32 nbatch; /* USER CODE END PTD */ /* Private define ------------------------------------------------------------*/ /* USER CODE BEGIN PD */ /* USER CODE END PD */ /* Private macro -------------------------------------------------------------*/ /* USER CODE BEGIN PM */ /* USER CODE END PM */ /* Private variables ---------------------------------------------------------*/ /* USER CODE BEGIN PV */ /* USER CODE END PV */ /* Private function pro

Hi,thereI'm a new learner and I'm trying to deploy an AI example given by ST official ,I trained the model using example dataset and generated an MDK file.I see there has some APIs for users to create and init neural net work like these:AI_API_ENTRYai_error ai_network_create_and_init(ai_handle* network, const ai_handle activations[ ], const ai_handle weights[ ]);And my questions :Where can I find "network" handle ,"activations" and "weights"?Or does any one can offer me your project as an example?Much thanks.

Hi,I'm following introductory course on Embedded AI, the course used Cube-Mx AI  3.4.0 AI Package, which generated a 'Data Processor .c' file. we need the file to insert user code for our application. But no where in the file structure in the up-to-date 9.0.0 could be found. Please advise if there is equivalent file in the generated code. Best regards  

Hi all,Based on the tutorials, I am trying to create an AI lib using NanoEdgeAI.The lib seems created successfully, see attachment.But when I add it to the MDK project and compiler it, some errors like below is present:When I remove the libneai.a, the project can be compiler normally.Could you tell me what I was wrong?Thanks.