Edge AI

How can Edge AI help your business?

Works independently of internet or cloud access – reliable even in hard-to-reach locations.
Uses less energy thanks to low-power components, extending the device’s lifespan.
Increases data security by processing it locally on the device.
Reacts instantly to critical events, without delays caused by sending data to the cloud.
Scales without additional cloud infrastructure or data transfer costs.

Real-life use cases utilizing Edge AI

By choosing proper SoM or microcontroller with AI capabilities, the device is able to perform given tasks on the edge, without the need of cloud processing and data transfer consumption. Here is the list of the most common use cases:

Predictive maintenance & anomaly detection (vibration, ultrasounds)

Sounds recognition (cry, siren, dog barking)

Distinguish music from the background noise

People counting using thermal or RGB cameras

Crash detection on wearables

Automated quality control

Objects detection and recognition on video stream

Smart voice assistant

Our hardware

STM32N6570-DK

For projects that need more computing power than standard microcontrollers can offer, we use STM's flagship development kit, based on the STM32N6 with a dedicated ST Neural-ART AI accelerator. The included Sony camera and integrated display significantly reduce the time needed to build and test solutions - for example in computer vision.

Raspberry Pi

Despite lacking a built-in AI accelerator, we regularly use Raspberry Pi in our solutions thanks to its Linux system and rich set of interfaces for cameras, microphones, and sensors. Model optimization and quantization allow us to run real-time image and audio processing on it.

ESP32-C3

ESP is well-known for its quality of microcontrollers and now it offers also running AI and ML models on the device. Cost and size-optimized, perfect for tasks like sounds recognition or predictive maintenance.

Infineon PSOC™ 6 AI Eval Kit

Small size, low-power and quite powerful. Perfect for sounds recognition, predictive maintenance and many other tasks.

JetNa

Utilizing Nvidia Jetson which is one of the most powerful Edge AI modules on the market. Perfect for heavier tasks like real-time video processing, objects tracking, behavior recognition and others.