Silicon Labs and Edge Impulse look to accelerate ML applications

1 min read

Silicon Labs and Edge Impulse, a development platform for machine learning on edge devices, are collaborating on the rapid development and deployment of machine learning (ML) on Silicon Labs EFR32 wireless SoCs and EFM32 microcontrollers (MCUs).

Implementation of the Edge Impulse tool will enable complex motion detection, sound recognition and image classification on low-power, memory-constrained, and remote edge devices.

Studies have shown that 87% of data science projects never reach full production, often due to artificial intelligence/ML implementation challenges. This new collaboration between Silicon Labs and Edge Impulse is intended to enable device developers to generate and export the ML models directly to the device or Simplicity Studio, the integrated development environment from Silicon Labs, with the click of a button, implementing machine learning in minutes.

“Silicon Labs believes the infusion of machine learning into the edge devices we help create will make the IoT smarter,” said Matt Saunders, vice president of IoT at Silicon Labs. “The secure, private and user-friendly tool from Edge Impulse saves developers time and money when implementing machine learning and enables amazing new user experiences across real-world commercial applications, from predictive maintenance to asset tracking to monitoring and human detection.”

Edge Impulse allows developers to quickly create neural networks across a wide range of Silicon Labs products for free, with integrated deployment to Simplicity Studio. By embedding state-of-the-art TinyML models on EFR32 and EFM32 devices such as MG12, MG21 and GG11, the solution enables:

• Machine learning
• Real-world sensor data collection and storage
• Advanced signal processing and data feature extraction
• Deep Neural Network (DNN) model training
• Deployment of optimised embedded code

The Edge Impulse tool also leverages Edge Impulse’s Edge Optimized Neural (EON) technology to optimise memory use and inference time.

"The industrial, enterprise and consumer applications of embedded ML are endless,” said Zach Shelby, co-founder and CEO of Edge Impulse. “Integrating ML with the advanced development tools and multi-protocol solutions from Silicon Labs unlocks robust wireless development opportunities for customers.”

Edge Impulse support is now available for the Silicon Labs Thunderboard Sense 2 and Silicon Labs wireless SoCs and MCUs