Driving Innovation in AI

Plumerai showcases computer vision solutions on small, embedded devices to address use cases that include people detection and face identification. The platform is built on the high-performance Renesas RA8x1 MCU with Arm Cortex-M85 and Arm Helium vector extensions to accelerate people-detection neural networks using just 300 KB RAM.

Himax WiseEye 2 runs vision-based AI edge models, including face detection, human pose estimation, and object detection. Its heterogeneous solution combines Arm Cortex-M55 with Arm Helium technology, Arm Ethos-U55 NPU, and support for TensorFlow Lite Micro and TVM to enable high-speed vision processing.

Sensory showcases its speech recognition and natural language understanding platform, built on Arm Cortex-M55 and the Arm Ethos-U55 NPU. Partitioning workloads across the Cortex-M and Ethos-U processors, Sensory enabled advanced wake words, voice control, and NLU capabilities, while meeting the tight memory and power limits of consumer devices.

There are many challenges involved in deploying AI speech recognition. Beijing-based SoundPlus Technologies worked with Arm to optimize a speech enhancement algorithm to run on Arm’s Cortex-M4F and Cortex-M33 MCU, achieving a balance between power consumption and performance.

Ignitarium’s real-time noise suppression software uses machine learning to suppress background noise. This exciting step in the innovation of audio and video segments designed for low-cost edge devices is enabled by the power efficiency and mature software support for machine learning functions found in Arm processors.

Rainforest Connection uses Arm-powered CPUs and machine learning technologies to create acoustic monitoring systems called Guardians that help protect fragile rainforest ecosystems. Since the data they collect can be processed in both the cloud and on the endpoint, Guardians are energy-efficient, self-contained, and adaptable.

Analysts predict the healthcare-related IoT market will exceed $10 billion by 2024. Arm is enabling this growth with machine learning in low-power devices. Quvium is a medical device company whose Sonasure cough monitor leverages Arm processors’ machine learning capabilities for identification and analysis of patient baselines.

Respiro is an inhaler that acts as a powerful tool in the treatment of respiratory conditions. It uses Arm-powered machine learning sensors to track and measure a patient’s breath, capturing personalized data to provide insight on how to better manage the condition. These machine learning algorithms can be applied across many disciplines.