Overview of image classification model by Lobe
Lobe is a free, easy-to-use Microsoft desktop application that allows you to build, manage, and use custom machine learning models. With Lobe, you can create an image classification model to categorize images into labels that represent their content. You can directly upload Lobe models to AI Builder for use with Power Apps and Power Automate.
Here's a summary of how to prepare a model in Lobe:
Import and label images.
Train your model.
Evaluate training results.
Play with your model to experiment with different scenarios.
Export and use your model in an app.
If you've previously used the AI Builder custom object detection model, you're already familiar with Lobe.
Below are some examples of how Lobe can take your machine learning ideas to production with Power Apps and Power Automate:
Identify defects in a manufacturing line.
Alert you when a shelf is empty.
Identify signs in your environment.
Filter or verify user-submitted images based on content.
Notify you when a shipment arrives.
Work with online retail images.
Provide information about images in the field.
What is image classification?
Image classification is the categorization of images into labels that represent their content overall. Image classification models learn to find patterns such as textures, colors, and shapes from your images. These patterns can then be used to identify your labels.
For example, a project named Contoso Insurance Claims might have labels named Dent, Scratch, and Broken Glass to indicate types of damage to vehicles. A claims adjuster might use these images to determine the insurance company's liability.
With image classification by Lobe, you train all your models on your computer. When you train this way, you don't use AI Builder credits. Lobe is integrated with Microsoft Power Platform, which adds the ability to export your model to AI Builder for use in Power Apps and Power Automate.
Select AI Builder > Explore > Images > Image classification.
On the next page, select Download Lobe to download and install the Lobe app on your computer.
Use image classification by Lobe
Within Lobe, create a new project by selecting Import.
Import and label your images, and train the classification model. To learn how to do this, go to the Lobe Help site.
(Optional) Tour Lobe and better understand image classification by watching the video on the Lobe home page.
Consider image quantity and object balance
You'll want to upload enough images to train your model accurately, and make sure there are enough images to represent each object. This will increase consistency when training your model.
For tips on choosing enough images, go to Data quantity and data balance.
Adjust your model architecture
Lobe determines the best model architecture for your project, but you have the option to change it. Changing your project will reset any training that's been completed and automatically train a new model.
The following are recommendations for Microsoft Power Platform users:
Speed: Use this setting as default for apps or flows that require quick response times.
Accuracy: Use this setting for apps or flows that batch jobs or have slower processing requirements.
Submit and view feedback for