Explore real-world edge
ML projects

Optimized person detection using the Object Detection transfer learning block which fine tunes a pre-trained object detection model on your data. It gives a good performance even with relatively small image datasets. This object detection learning block relies on MobileNetV2 SSD FPN-Lite 320x320.

A project to predict the anomalous operation in Industrial machineries like Pump, valves & fans with an Arduino Portenta.

Here we have 3 types of data required to clasify and predict fan noise of 3D printer [Nearest industrial machine available right now]

Estimate rotor temperature to assist in building better motors. A precise torque estimate leads to more accurate and adequate control of the motor, reducing power losses and eventually heat build-up.

Classify fan operation, whether it is currently off, working normally, or in anomalous way.

Pnemonia detection can still be out of reach for many in underpriveleged areas. Traditional detection of pneumonia is achieved by using x-rays, chest scans or extracting blood serum in extreme cases. However, this requires time, resources, skilled personnel, and it is expensive. This example demonstrates how to use Edge Impulse to train a computer vision model and its hardware partners for providing early warning detection of potential pneumonia.

Arn atificial nose can be used for a wide variety of applications, from helping people suffering from anosmia to spot the smell of burning food or spoiled milk, to monitoring the cleanliness of office buildings, or detect the smell of smoke.

Fast new object detection on constrained devices, FOMO (Faster Objects More Objects).

Detect anomalies during vaccine transportation.

Machine motion monitor of a toy tower crane with an Arduino Nano 33 BLE Sense.

Cold Chain Monitoring is a multi-billion-dollar market that spans across many industries such as transportation and logistics, grocery retail, and smart agriculture. One of the key sensing metrics that makes applications in this space successful is the monitoring of environment temperature where perishable products are produced, processed, packaged, stored, transported, and sold. In this example we show you how to use Edge Impulse for deploying embedded machine learning on low-power temperature sensors that can help warn the system owner ahead of time before products such as food and vaccines have exceeded their temperature storage limits.

This example demonstrates how to use CO2 and VOC measurements using the Himax WE-I Plus board and the combo CCS811+BME280 sensor board for creating a system that can be applied to provide early warning detection of hazardous air quality situations in workplace, factory, and manufacturing plant environments.

By using temperature sensor reading patterns and Edge Impulse's regression learning block, this example demonstrates an early warning detection system that informs the user how close the system is to reaching its critical temperature point, something that should be avoided at all costs in most industrial situations.

Object classification based on their radar signature. This project aims to classify four different classes: person, uav, bicycle, and vehicle.

Detect wether a faucet is on or not with the use of sound.

Lacuna Space, Edge Impulse, a microcontroller and LoraWAN can promote the conservation of endangered bird species by monitoring bird calls in remote areas. This example takes you through the process of deploying a bird sound classification project using an Arduino Nano 33 BLE Sense board and a Lacuna Space LS200 development kit. The inferencing results are then sent to a TTN (The Things Network) application.

Project to detect different types of AC motor faults based on 48 statistical features.

Detect anomalies in the state of operation of a converyor belt.

Detect person to wake up a system in idle mode.

Compact classifier that identifies a ball bearing fault in a two-phase AC motor.

Count different objects on a conveyor belt.

Project to classify the 10 Google Speech Commands.

Estimate the weight of rice by analizing an image of the rice stack.

Estimate different boxing moves with an accelerometer.

This is the finished Edge Impulse project for the tutorial 'Adding sight to your sensors'.

Tea or chocolate identification using TVOC sensor.

Detect bearing failure, axis inclination, and shaft misalignment faults for two phase AC motors.

Learn how to detect that the wearing of a helmet is well respected by all workers using a machine learning object detection model.

Digits recognition has multiple autonomous applications in grocery retail (price label mismatch / use by date inspection), manufacturing (production line inspection, quality control monitoring), utility metering (meter reading of smart and legacy meters) and adminstration (license plate detection / handwriting recognition), among others. In this example we show how to leverage low power computer vision with Edge Impulse and the Arduino Portenta H7 + Vision Shield and create a price label detection system using digits recognition.

This example uses the object detection learning block in Edge Impulse to identify products in your shopping basket (a la a smart cart) or be applied identifying out of stock items on retail shelves to help streamline stock replenishment. This can be deployed on an embedded Linux-based system such as the Raspberrry Pi 4 or NVIDIA Jetson Nano and a standard USB camera.

Project to detect wounded animals in wild using Kria KV260.

Teach a camera to detect the state of an egg.

Visually detect fires.

Gesture recognition is one of the most popular human to machine interface (HMI) techniques in consumer, security, and home automation products. Using Edge Impulse and the Infineon Sense2Gol Pulse kit's doppler radar capabilites, this example demonstrates how to run radar gesture recognition models on embedded devices with Doppler radar sensors using minimal device resources.