An integrated tool designed for engineers to quickly implement AI models on boards
As AI technology accelerates its integration into various embedded systems, deploying trained AI models onto resource-constrained microcontrollers has become a significant challenge for engineers. To assist developers in rapidly implementing AI solutions, Nuvoton Technology has launched the NuML Toolkit, a tool that integrates model conversion, project generation, deployment, and debugging processes, specifically optimized for the NuMicro® M55M1 microcontroller platform, receiving widespread acclaim from users.
The NuML Toolkit provides comprehensive process assistance, including:
-
Model Conversion and Deployment: Automatically converts Full-INT8 quantized models into .cpp files and integrates them into Arm® Keil® project templates
-
Dual Toolchain Support: Supports both Keil (μVision®5 / Arm Compiler 6) and GNU Compiler Collection (make / gcc)
-
High Integration: Automatically opens the Keil project and initiates programming after deployment, allowing immediate review of inference results
-
Flexible Scalability: Developers can adjust details such as Tensor Arena, input/output processing according to model requirements
-
Cross-Platform Support: Uses Miniforge to set up the environment, simplifying dependency management and quickly importing toolchains
The three key advantages of using the NuML Toolkit are:
-
Simple and Fast Model Conversion
Through command line operations, users only need a few parameters to automatically generate the corresponding AI MCU project without manually writing complex program code.
-
Complete Process IntegrationFrom model quantization → project building → microcontroller programming → result verification, the fully integrated support allows both beginners and experienced engineers to use it with confidence.
-
High Flexibility and MaintainabilityThe generated Keil project includes a modular program code architecture (such as Application, Model, Device, etc.), making it easy for customers to modify and maintain according to actual needs. Model management is flexible, and large models can choose to use SD cards or HyperRAM™, automatically switching memory usage strategies based on resource allocation.
The NuML Toolkit is a highly efficient tool tailored for microcontroller AI applications, assisting customers in completing AI model deployment and verification in the shortest time possible. It has been widely applied in various scenarios such as smart lamps, gesture recognition, voice recognition, and image detection, becoming one of the preferred tools in the Nuvoton NuMicro® AI MCU developer community.
For further downloads of the tool or to obtain educational resources, please contact the Nuvoton team through the Nuvoton AI website www.nuvoton.com/ai for more detailed information.
END
