# featureExtractor The underlying base model the training session uses to extract image features as it trains an image classifier. ``` var featureExtractor: MLImageClassifier.FeatureExtractorType { get set } ``` ## Discussion When you train an image classifier with Create ML, you don’t train a complete model that converts images to labels. Instead, Create ML leverages a *feature extractor*, which is another model that identifies a set of general but distinguishing image characteristics. You can either use Create ML’s `scenePrint` feature extractor or provide your own custom feature extractor. In either case, you train your model to map the feature extractor’s output to labels in a process known as *transfer learning*. Your model effectively borrows the knowledge of the feature extractor to accelerate its own training process while requiring fewer training images. --- Copyright © 2026 Apple Inc. All rights reserved. | [Terms of Use](https://www.apple.com/legal/internet-services/terms/site.html) | [Privacy Policy](https://www.apple.com/privacy/privacy-policy)