Amazon Rekognition Custom Labels now supports single object training
Amazon Rekognition Custom Labels is an automated machine learning (ML) feature that enables customers to quickly train their own custom models for detecting business-specific objects and scenes from images - no ML experience is required. For example, customers train a custom model to find their company logos in social media posts, identify their products on store shelves, or classify unique machine parts in an assembly line. Starting today, Amazon Rekognition Custom Labels now allows customers to train object detection projects for a single object (label).
Customers told us that for certain object detection use cases, they only need to find a single class of object to determine its presence or absence. To train a custom model with Amazon Rekognition Custom Labels, customers currently need to provide a minimum of two objects (labels). That means customers have to create a second object (either another object or a ‘non-object’) label. With this new feature, customers no longer have to create a second label for object detection use cases and can simply train a model using the single object label they care about. This feature is now available in all Amazon Rekognition Custom Labels regions. You can find the list of supported regions in the region table.
For more details on how to label images for use with Amazon Rekognition Custom Labels, please refer to the feature documentation. To learn more about Amazon Rekognition Custom Labels, please visit the product webpage.