ST AIoT Craft NEW
An online tool to fast-track the development of sensor-to-cloud solutions leveraging ST components with in-sensor AI.
Tools
Create, optimize, and deploy your machine learning algorithms.
- Create sensor-to-cloud solutions with ST boards
- Program the machine learning core to run in-sensor AI
- Explore end-to-end project examples.
ST AIoT Craft NEW
1.0.0
An online tool to fast-track the development of sensor-to-cloud solutions leveraging ST components with in-sensor AI.
- Create sensor-to-cloud solutions with ST boards
- Program the machine learning core to run in-sensor AI
- Explore end-to-end project examples.
- Create sensor-to-cloud solutions with ST boards
Use the online tool to streamline the development of AI-enabled IoT nodes.
- Program the machine learning core to run in-sensor AI
Quickly create decision tree algorithms running on the machine learning core embedded in MEMS sensors.
- Explore end-to-end project examples.
Discover proof-of-concepts using ST reference designs.
NanoEdge AI Studio
A free AutoML software for adding edge AI to embedded projects, guiding users step by step to easily find the optimal AI model for their requirements.
- AutoML: automatic machine learning model generator
- Supports more than 1,000 Arm® Cortex®-M microcontrollers
- User-friendly software for end-to-end edge AI deployment
NanoEdge AI Studio
4.6
A free AutoML software for adding edge AI to embedded projects, guiding users step by step to easily find the optimal AI model for their requirements.
- AutoML: automatic machine learning model generator
- Supports more than 1,000 Arm® Cortex®-M microcontrollers
- User-friendly software for end-to-end edge AI deployment
- On-device learning for optimized anomaly detection
- AutoML: automatic machine learning model generator
NanoEdge AI Studio selects the best ML algorithm for a given microcontroller (low code / no code solution).
- Supports more than 1,000 Arm® Cortex®-M microcontrollers
Supports all STM32 microcontrollers, more than 100 ST development boards, 20 Arduino boards, and over 1,000 Arm® Cortex®-M microcontrollers.
- User-friendly software for end-to-end edge AI deployment
No expertize required. Create tiny, state-of-the-art AI models for microcontrollers in record time.
- On-device learning for optimized anomaly detection
NanoEdgeAI Studio offers a unique and patented approach that enables on-device learning directly on the microcontroller.
ST Edge AI Developer Cloud
A free online platform to easily optimize and benchmark edge AI models across a variety of ST devices. It relies on the ST Edge AI Core to perform AI model optimizations and validations.
- Online AI benchmark service on multiple ST devices
- Optimize and profile your AI model
- Run your AI models on the ST board farm
ST Edge AI Developer Cloud
A free online platform to easily optimize and benchmark edge AI models across a variety of ST devices. It relies on the ST Edge AI Core to perform AI model optimizations and validations.
- Online AI benchmark service on multiple ST devices
- Optimize and profile your AI model
- Run your AI models on the ST board farm
- Online AI benchmark service on multiple ST devices
Evaluate AI model performance in the cloud based on ST devices.
- Optimize and profile your AI model
Execute ST Edge AI Core technology to optimize your AI model and get insights into the model execution and performance on ST devices.
- Run your AI models on the ST board farm
Access the ST board farm and enjoy real-time access to physical ST products remotely. Review the performance of your AI models for the selected devices.
MEMS Studio
A complete software solution for desktops to enable edge AI features on smart sensors. It allows users to analyze data, evaluate embedded libraries, and design no-code algorithms for the entire portfolio of MEMS sensors.
- Collect, label, and analyze sensor data
- Profile and optimize NN and ML models for the ISPU
- Configure the MEMS machine learning core
MEMS Studio
1.3.0
A complete software solution for desktops to enable edge AI features on smart sensors. It allows users to analyze data, evaluate embedded libraries, and design no-code algorithms for the entire portfolio of MEMS sensors.
- Collect, label, and analyze sensor data
- Profile and optimize NN and ML models for the ISPU
- Configure the MEMS machine learning core
- Select the decision tree features and filters
- Collect, label, and analyze sensor data
Line charts of MEMS sensor data.
- Profile and optimize NN and ML models for the ISPU
Load, analyze, and evaluate models from Keras, ONNX, and TFLite on the intelligent sensor processing unit (ISPU).
- Configure the MEMS machine learning core
Collect data, generate decision trees, and configure sensors with a machine learning core (MLC).
- Select the decision tree features and filters
Use the automatic filtering and feature selection for a simpler MLC configuration process.
ST Edge AI Core
A command-line interface (CLI) tool to optimize and compile edge AI models for multiple ST devices, including microcontrollers, microprocessors, and MEMS sensors.
- Import AI models from the most popular ML frameworks
- Review the detailed analysis of the selected algorithms
- Validate and optimize the models for different devices
ST Edge AI Core
2.0.0
A command-line interface (CLI) tool to optimize and compile edge AI models for multiple ST devices, including microcontrollers, microprocessors, and MEMS sensors.
- Import AI models from the most popular ML frameworks
- Review the detailed analysis of the selected algorithms
- Validate and optimize the models for different devices
- Generate executable code for the selected ST hardware
- Import AI models from the most popular ML frameworks
Example of the installer.
- Review the detailed analysis of the selected algorithms
- Validate and optimize the models for different devices
- Generate executable code for the selected ST hardware
ST Edge AI Model Zoo
A collection of reference edge AI models optimized to run on ST devices with associated deployment scripts. The model zoo is a valuable resource to add edge AI capabilities to embedded applications.
- Choose from a collection of AI models optimized for ST devices
- Retrain any model from datasets using the scripts
- Deploy AI models on the application using the scripts
ST Edge AI Model Zoo
A collection of reference edge AI models optimized to run on ST devices with associated deployment scripts. The model zoo is a valuable resource to add edge AI capabilities to embedded applications.
- Choose from a collection of AI models optimized for ST devices
- Retrain any model from datasets using the scripts
- Deploy AI models on the application using the scripts
- Bring your own model (BYOM) or bring your own data (BYOD)
- Choose from a collection of AI models optimized for ST devices
The neural network models provided in the model zoo are optimized for various applications on the ST target devices.
- Retrain any model from datasets using the scripts
- Deploy AI models on the application using the scripts
- Bring your own model (BYOM) or bring your own data (BYOD)
STM32Cube.AI
A free STM32Cube expansion package, X-CUBE-AI allows developers to convert pretrained edge AI algorithms automatically, such as neural network and machine learning models, into optimized C code for STM32.
- Optimize NN and ML models for STM32 platforms
- Profile NN and ML models on STM32 platforms
- Evaluate NN and ML models on STM32 platforms
STM32Cube.AI
9.0.0
A free STM32Cube expansion package, X-CUBE-AI allows developers to convert pretrained edge AI algorithms automatically, such as neural network and machine learning models, into optimized C code for STM32.
- Optimize NN and ML models for STM32 platforms
- Profile NN and ML models on STM32 platforms
- Evaluate NN and ML models on STM32 platforms
- Available as an STM32CubeMX expansion package
- Optimize NN and ML models for STM32 platforms
Import your own neural network models into STM32CubeMX, select optimization options, and generate the optimized C code corresponding to the input models.
- Profile NN and ML models on STM32 platforms
X-CUBE-AI analyzes the NN model and generates a profiling report that details the NN memory requirements and the inference time, both for the complete network and for each layer.
- Evaluate NN and ML models on STM32 platforms
- Available as an STM32CubeMX expansion package
High Speed Datalog
This tool allows users to manage the acquisition and labelling of sensor datasets. Datasets are stored onto microSD cards or streamed via USB. The data acquisition can be controlled using the graphical user interface (GUI), the command line interface (CLI), or Bluetooth with a smartphone.
- Capture and monitor high-rate data
- Acquire and manage data with the Python SDK
- Port your project across multiple MCU series
High Speed Datalog
2.1.1
This tool allows users to manage the acquisition and labelling of sensor datasets. Datasets are stored onto microSD cards or streamed via USB. The data acquisition can be controlled using the graphical user interface (GUI), the command line interface (CLI), or Bluetooth with a smartphone.
- Capture and monitor high-rate data
- Acquire and manage data with the Python SDK
- Port your project across multiple MCU series
- Explore modular firmware examples
- Capture and monitor high-rate data
Monitor data acquisition and track live data on the user interface.
- Acquire and manage data with the Python SDK
Customize widgets for accurate data source management and visualization.
- Port your project across multiple MCU series
- Explore modular firmware examples
StellarStudioAI
The StellarStudio's AI plug-in for Stellar electrification (E) microcontrollers simplifies the development of neural networks in automotive systems, offering automatic model conversion, execution, and validation.
- Convert your AI models
- Create and review neural network performance reports
- Convert pretrained neural networks automatically
StellarStudioAI
1.2.0
The StellarStudio's AI plug-in for Stellar electrification (E) microcontrollers simplifies the development of neural networks in automotive systems, offering automatic model conversion, execution, and validation.
- Convert your AI models
- Create and review neural network performance reports
- Convert pretrained neural networks automatically
- Support for Keras, TensorFlow lite, ONNX
- Convert your AI models
Explore the GUI interface with Keras model generation.
- Create and review neural network performance reports
Use the IDE panel with validation and performance output.
- Convert pretrained neural networks automatically
Generate pretrained neural networks and convert them into efficient Ansi C libraries, which can be easily compiled, installed, and executed on Stellar E MCUs.
- Support for Keras, TensorFlow lite, ONNX
Integration of AI plug-in with the Stellar E development environment.
AI for OpenSTLinux
X-LINUX-AI is an STM32 MPU OpenSTLinux expansion package for running edge AI models on STM32MP1 and STM32MP2 microprocessors. It contains Linux® AI frameworks, as well as application examples.
- Pose estimation (Yolov8n)
- Semantic segmentation (DeepLabv3)
- Image classification (MobileNetv2)
AI for OpenSTLinux
5.1.0
X-LINUX-AI is an STM32 MPU OpenSTLinux expansion package for running edge AI models on STM32MP1 and STM32MP2 microprocessors. It contains Linux® AI frameworks, as well as application examples.
- Pose estimation (Yolov8n)
- Semantic segmentation (DeepLabv3)
- Image classification (MobileNetv2)
- Object detection (SSD MobileNetv2)
- Pose estimation (Yolov8n)
A collection of code examples for the most common computer vision applications.
- Semantic segmentation (DeepLabv3)
- Image classification (MobileNetv2)
- Object detection (SSD MobileNetv2)
Hand posture ToF AI
The hand posture recognition solution detects a set of hand postures based on an ST multizone Time-of-Flight sensor, eliminating the need for a camera.
- Several possibilities to create dataset
- Seven hand postures recognized in the proposed dataset
- Based on an 8 x 8 ranging distance and signal rate
Hand posture ToF AI
The hand posture recognition solution detects a set of hand postures based on an ST multizone Time-of-Flight sensor, eliminating the need for a camera.
- Several possibilities to create dataset
- Seven hand postures recognized in the proposed dataset
- Based on an 8 x 8 ranging distance and signal rate
- Deployed on an STM32 microcontroller
- Several possibilities to create dataset
Start with your own dataset or select an existing dataset.
- Seven hand postures recognized in the proposed dataset
The training process on ST or on custom datasets shows the related accuracy/loss charts and confusion matrix.
- Based on an 8 x 8 ranging distance and signal rate
- Deployed on an STM32 microcontroller
