STM32Cube.AI Upgrade Version Released
Rapid design of AI/ML solutions with minimal investment

STMicroelectronics (hereinafter ST) has introduced an AI model optimization tool for developers who need to use neural networks.

ST announced on the 16th that it has released upgraded versions of NanoEdge AI Studio and STM32Cube.AI, improving its tools to accelerate embedded AI and Machine-Learning (ML) development projects.

With this tool, AI and ML can be easily moved to edge applications, and benefits such as design-based privacy, deterministic and real-time response, excellent reliability, and low power consumption can be realized at the edge through AI/ML.

NanoEdge AI Studio is an automated ML tool that supports applications that do not require neural network development.

This tool can be used with MEMS sensors containing ST's proprietary Embedded Intelligent Sensor Processing Unit (ISPU) and STM32 microcontrollers (MCUs).

STM32Cube.AI is an AI model optimization tool and compiler for STM32 for developers who need to use neural networks.

The two new tools announced this time enable the rapid design and implementation of high-performance AI/ML solutions with minimal investment.It provides the ability to do so.

NanoEdge AI Studio version 3.2 includes an automatic data logger generator that increases development productivity. It supports the input of developer-defined sensor parameters, such as data transmission rate, range, sample size, and number of axes, including ST development boards.

Developers can generate binaries for development boards without writing code using these NanoEdge AI Studio features.

NanoEdge AI Studio provides new data manipulation features to help maintain dataset quality, which has a direct impact on machine learning performance, enabling users to clean and optimize data captured in NanoEdge AI Studio with just a few clicks.

A new validation step has also been added, allowing users to evaluate algorithms by checking general performance metrics such as inference time, memory usage, and accuracy, as well as F1 scores.

It also highlights more information about preprocessing and ML models related to the selected library.

The latest improvements to NanoEdge AI Studio include performance-enhancing regression algorithms and more ML models and pre-processing techniques for anomaly detection.

In addition, this tool supports the creation of smart libraries capable of predicting future system states using multiple regression models.

STM32Cube.AI version 7.3 is an essential tool for developing cutting-edge AI/ML solutions.

It is fully integrated with the STM32 ecosystem and can convert pre-trained neural networks into optimized C code for the industry's most widely used 32-bit Arm® Cortex® core-based MCU family.

The newly improved STM32Cube.AI enables more flexible optimization of neural networks.

This tool can tune existing neural networks to meet performance requirements or modify them to fit within limited memory, and these You can utilize the best features through balanced optimization.

The new updated version also provides support for TensorFlow 2.10 models and new kernel performance improvements.