Edge Impulse is excited to announce its direct integration with the Arm® Keil® MDK.
Keil MDK is a toolkit designed for developers working with Arm microcontrollers, offering a set of software tools that help them write, organize, debug, and optimize their code efficiently for Arm chips.
This integration now provides thousands of embedded and firmware Arm developers and engineers with easy access to cutting-edge AI and machine learning models, simplifying and accelerating collaborations across teams within the Edge Impulse platform, and speeding up time to market for edge AI products.
Learn more about the integration in the press release.
In this blog post, we will take a deeper dive into the core benefits to developers and provide access to technical documentation.
Leveraging Open-CMSIS-Pack for seamless integration
Adopted by both Edge Impulse and Arm, the open Common Microcontroller Software Interface Standard (CMSIS) makes this integration possible.
The Open-CMSIS-Pack project standardizes the packaging and distribution of software components. It offers tools for the creation, management, and deployment of software packs, supporting over 10,000 microcontrollers and enhancing the reusability and compatibility of software components across the edge AI, IoT and embedded systems domains. In other words, you’ll be able to build solutions efficiently at a production-ready level that are above and beyond what's currently available in the marketplace.
And as an added benefit from the adoption of Open-CMSIS-Pack, Edge Impulse users have hardware access beyond Arm Keil MDK, to any development platform that conforms to the Open-CMSIS standard! Learn more about the various platforms that conform to the standard.
Boost performance of your Arm-based edge AI solution
The market-leading Edge Optimized Neural (EON) Compiler, created by Edge Impulse, runs models with reduced RAM and flash usage, all the while maintaining accuracy comparable to TensorFlow Lite for Microcontrollers. It significantly reduces device resource utilization and saves inference time.
When combined with the Arm Compiler, you can expect a dramatic reduction in footprint, and the ability to bring production-grade performance to your Arm-based edge AI solutions. You will see:
- Up to 70% less RAM usage in addition to the saving from the Arm Compiler
- Up to 40% less flash usage in addition to the saving from the Arm Compiler
- The same accuracy as TensorFlow Lite for Microcontrollers
- Faster inference times
- Reduced overall power consumption of your edge AI applications
Maximize savings while minimizing BOM: EON Tuner
Separately, the EON Tuner, our award-winning AutoML tool, empowers you to keep costs to the minimum requirement for your given edge AI solution. By performing end-to-end optimizations, from the digital signal processing (DSP) algorithm to the machine learning model, you’ll find the ideal trade-off between these two blocks to achieve optimal performance on your target hardware.
For full details, please refer to our EON Tuner documentation.
How to get started
Whether you’re already utilizing the Arm platform, or simply interested in getting started, we’ll help you along the journey. The integration supports a wide range of development environments, including CLI and GUI, across macOS, Linux, and Windows. The CMSIS pack manager guarantees your SDK remains up-to-date with the latest releases, ensuring your projects benefit from the latest enhancements and support.
The process is simple:
- Connect your chosen Arm development platform and pull the latest Edge Impulse SDK pack
- Deploy your Edge Impulse project as a CMSIS pack and import it into your project
Access our Arm Keil Documentation here.
Propel your projects forward with the Arm Keil MDK and Edge Impulse
By seamlessly integrating Edge Impulse across the Arm Keil MDK ecosystem, developers gain the significant advantage of leveraging both the Arm Compiler and Edge Impulse EON Compiler, resulting in increased speed, efficiency, and better compatibility and flexibility with your projects.
From concept to deployment, the journey to your edge AI ambition just got easier and faster. To get started on your next project, access the documentation guide. We can't wait to see what you build.