Socks Classification with Arduino UNO Q and Edge Impulse

Webcam, Logitech® HD Pro

Arduino UNO Q

Edge Impulse Studio

<html><body>Believe me! Matching socks is a nightmare, so having an intelligent system saying if a sock matches with another sock is the best weekend project ever. We are going to create a project on the Arduino UNO Q that detects when the socks are paired or not using the Edge Impulse machine learning platform. <h3>Hardware required</h3>To build this project, you’ll need the following: <ol><li>Arduino UNO Q </li><li>USB Webcam </li><li>USB-C Hub/Power Supply (since the UNO Q uses USB-C for power and data, a hub is needed to connect the webcam and power simultaneously) </li><li>Wi-Fi connection </li></ol><h3>Software setup</h3>For the software, you’ll need: <ol><li>An Edge Impulse studio account ( <a href="https://studio.edgeimpulse.com/signup?utm_medium=partner&amp;utm_source=none&amp;utm_campaign=21157154-new-user-acquisition_fall2025&amp;utm_content=arduino-project-hub-socks-classification">sign up here it's free </a>) </li><li>Terminal access to your Arduino UNO Q (via SSH or Serial). </li></ol> <h3>Tutorial</h3>First of all, Edge Impulse is a platform that enables you to collect data and generate Machine Learning trained models and deploy them on embedded hardware. From microcontrollers such as Arduino Nicla Vision or Linux-based boards like the Arduino UNO Q. That means that no GPU usage is required locally for training a machine learning model because all the neural network training is done beforehand in the Edge Impulse Studio with advanced methods. In this project, we will collect images from our socks from the Arduino UNO Q, add it to the Edge Impulse Studio, train an image classification model, and deploy it into the UNO Q. Then we will be able to classify images with the image classifier model from the stream captured via a USB camera connected to the Arduino UNO Q. From now, learn how to generate a Machine Learning model using Edge Impulse and deploy an image classification model in the Arduino UNO Q and run inference. By the end of this project, you’ll be able to reunite all of your unpaired socks together again (and try out other edge AI use cases)! Edge Impulse Studio Go to the Edge Impulse website and <a href="https://studio.edgeimpulse.com/signup?utm_medium=partner&amp;utm_source=none&amp;utm_campaign=21157154-new-user-acquisition_fall2025&amp;utm_content=arduino-project-hub-socks-classification">create an account </a>. Create a project, in this case Socks-Classification . <figure><img src="https://projects.arduinocontent.cc/64ab9b32-7242-463e-948f-998989767712.png"/><figcaption>Edge Impulse Studio - Project creation</figcaption></figure>Connect a device and start taking pictures Once the Edge Impulse project is created, select the project and start collecting the data to train the Machine Learning model. Plug the Arduino UNO Q to the USB Hub and plug it in your computer. We are going to use the <code class="code">Edge Impulse Linux CLI </code>to connect the device to the Edge Impulse project. Access via SSH to your Arduino UNO Q and install the Edge Impulse Linux needed libraries. Find more information about installing the Edge Impulse dependencies in the <a href="https://docs.edgeimpulse.com/hardware/boards/arduino-uno-q#installing-edge-impulse-dependencies">Edge Impulse documentation </a>. Type this to install the Edge Impulse dependencies in the Arduino UNO Q: <pre>sudo apt update </pre><pre>curl -sL https://deb.nodesource.com/setup_20.x | sudo bash - </pre><pre>sudo apt install -y gcc g++ make build-essential nodejs npm sox gstreamer1.0-tools gstreamer1.0-plugins-good gstreamer1.0-plugins-base gstreamer1.0-plugins-base-apps </pre><pre>sudo npm install edge-impulse-linux -g --unsafe-perm </pre>With the libraries installed successfully type: <pre>edge-impulse-linux </pre>This will start a wizard which will ask you to log in, and choose an Edge Impulse project. If you want to switch projects, run the command with <code class="code">--clean </code>. In case the camera or microphone are not detected successfully <a href="https://forum.edgeimpulse.com">contact Edge Impulse in the forum </a>. Alternatively navigate to Devices on the main menu and then click Connect a new device on the top right. A modal will pop up to connect a new device. For this project, it's possible to use your mobile phone to take pictures to train the model (it's the fastest way to build a dataset!). <figure><img src="https://projects.arduinocontent.cc/144441d8-6f16-4145-9e97-432aae1ad4db.png"/><figcaption>Edge Impulse Studio - Devices</figcaption></figure>At that point, in the <code class="code">Data Acquisition </code>section there will be the Camera Stream from the Arduino UNO Q camera ready to take samples. Click on Label and write <code class="code">Pair </code>, and start taking pictures of paired socks. Once you have taken 40-50 pictures of different paired socks (depending on your jungle of socks), change the label to <code class="code">NotAPair </code>and take pictures of your socks unpaired. <figure><img src="https://projects.arduinocontent.cc/60cb4150-b528-4988-97b1-d2213a78795d.png"/><figcaption>Edge Impulse Studio - Data Acquisition - Taking samples of images</figcaption></figure>There is the possibility to split automatically the pictures (80/20) as training pictures and testing pictures or you can do it manually. <h3>Train the model of the Project</h3>Create the impulse of the project Go on the main menu to <code class="code">Create impulse </code>. For this project, we are using Image data on a resolution of 96x96 . Click Add a processing block and add <code class="code">Image </code>. Click Add a learning block and add <code class="code">Classification </code>. Finally click Save Impulse. <figure><img src="https://projects.arduinocontent.cc/ab3f19d7-1b1e-4a3f-ae2f-0c9ca8902fbe.png"/><figcaption>Edge Impulse Studio - Create Impulse - Image Classification</figcaption></figure>Now there are more menus below Create impulse on the main menu. Click on <code class="code">Image </code>. And click on Save Parameters with RGB color depth. <figure><img src="https://projects.arduinocontent.cc/bb0e4462-5929-41d4-9c72-7215a2f44d88.png"/><figcaption>Edge Impulse Studio - Save Parameters</figcaption></figure>This sends us to <code class="code">Generate features </code>that will create a visualization of the dataset captured. With the visual model generated with the Training Data captured you can see how distinct the classified objects are. <figure><img src="https://projects.arduinocontent.cc/a14670b7-e493-4fbf-9971-91fdcb7c33fc.png"/><figcaption>Edge Impulse Studio - Generate Features</figcaption></figure>The data is processed; now we have to start training a neural network to recognize the patterns of the data. The neural networks are algorithms that are designed to recognize patterns. In this case, the neural network will be trained with image data as input, and it will try to map the images into one of the categories. <figure><img src="https://projects.arduinocontent.cc/78c8d7aa-2d3b-4957-af23-8ee3128c9b64.png"/><figcaption>Edge Impulse Studio - Classifier - Train teh Neural Network</figcaption></figure> Click Start training and the neural network will start to compute all the images and train to generate the Machine Learning model. After the model is done you’ll see accuracy numbers, a confusion matrix, and some predicted on-device performance on the bottom. You have now trained your model. <figure><img src="https://projects.arduinocontent.cc/a41c49cc-7298-403e-8b9d-42a138f25c28.png"/><figcaption>Edge Impulse Studio - Train Neural Network - Accuracy</figcaption></figure><h3> Test the model generated </h3>Now it’s time to deploy the model on the Arduino UNO Q and apply it to the real world. Since the Arduino UNO Q runs Linux, we don't need to compile a binary manually or flash firmware like a microcontroller. We can use the <code class="code">Edge Impulse Linux Runner</code>. In this project you don't need to click "Deployment" in the Edge Impulse Studio for this step; we will pull the model directly from the device using the CLI. Access the Arduino UNO Q via SSH. If you still have the Edge Impulse Linux CLI running, stop it (Control + C). Now, run the <code class="code">Edge Impulse Linux Runner</code> to automatically download your trained model and start the inference process. Type<pre>edge-impulse-linux-runner </pre><figure><img src="https://projects.arduinocontent.cc/8a0d467a-a4b9-41eb-9370-33d06f538fd5.png"/><figcaption></figcaption></figure>You should see output in your terminal indicating that the classification is running. However, to see the video stream, look for a line in the terminal that says:<pre>Want to see a feed of the camera and live classification? Go to http://&lt;IP_ADDRESS&gt;:4912 </pre>Open a browser on your computer (connected to the same WiFi network) and enter the IP address shown in the terminal with port 4912. <figure><img src="https://projects.arduinocontent.cc/8f514f72-b7c5-44f2-b498-337039c49fbc.png"/><figcaption></figcaption></figure><h3>Until next time </h3>We would love to hear from you to see what you are classifying with the Arduino UNO Q and Edge Impulse. We are always interested to see how the community puts the projects to work. Get in touch with us on our <a href="https://forum.edgeimpulse.com">Forums</a> and <a href="https://discord.gg/edgeimpulse">Discord</a> to show off your work or to ask questions. We will be more than happy to help. </body></html>

Socks Classification with Arduino UNO Q and Edge Impulse

Deploy an image classifier model on the Arduino UNO Q with a USB Camera. It’s AI to track your socks 🧦🧦!