SensiML Corporation announced it has added support for Silicon Labs’ new MG24 and BG24 Series 2 Bluetooth wireless SoCs within its SensiML Analytics Toolkit embedded development software. The MG24 and BG24 are noteworthy for their inclusion of a specialized AI accelerator and co-optimized TensorFlow library as part of these Matter-ready wireless SoCs for IoT edge applications.

SensiML provides embedded edge AI tools with the most complete end-to-end AI workflow addressing all aspects of AI development from advanced data collection and labeling to code generation and validation, and now extends its powerful yet easy-to-use AutoML software to these latest Silicon Labs AI accelerated SoCs. With support for the MG24 and BG24 AI accelerator, SensiML can deliver highly optimized performance up to 4x faster at one-sixth the power for applications that utilize the TensorFlow Lite neural network classifier engine. These attributes are particularly valuable for audio, high data-rate sensors, and battery-powered IoT sensor applications, which are becoming ubiquitous.

The SensiML Analytics Toolkit provides a complete solution for creating AI/ML-based designs targeted at intelligent IoT devices. The toolkit complements the AI acceleration capabilities built into the Silicon Labs MG24 and BG24 SoCs – allowing developers to quickly create sophisticated, low power intelligent IoT endpoints. SensiML’s tools also dramatically reduce development complexity by automating the design flow and optimizing the resulting firmware to deliver high-quality results while keeping the memory and power footprint as small as possible.

Applications that can particularly benefit from the combined SensiML and Silicon Labs solution include acoustic event detection, motion analysis, gesture and keyword recognition, anomaly detection, and predictive maintenance.

The latest version of the SensiML Analytics Toolkit includes optimizations to support the AI accelerator core in the SiLabs MG24 and BG24 Series 2 Bluetooth wireless SoCs as part of SensiML toolkit custom Knowledge Packs. These Knowledge Packs can be readily integrated into customized firmware applications and can be generated in the binary, library, and full C source code formats. Also included in the latest SensiML release is support for the Silicon Labs EFR32xG24 Dev Kit (xG24-DK2601B) boards allowing developers to rapidly evaluate a diverse array of sensor applications using the xG24 AI capabilities with SensiML on this sensor-packed small form factor IoT kit.