Enumeration Case

MLImageClassifier.FeatureExtractorType.scenePrint(revision:)

A feature extractor trained on millions of images.

Declaration

case scenePrint(revision: Int?)

Discussion

Because the scene print feature extractor is trained on millions of images of real world objects, it works best when extracting features from these kinds of images. It isn’t suitable for character recognition, for example, where the input images are highly binary in nature (pixels are either on or off).

When training an image classifier using scene print, or making predictions with the resulting model, be sure to use images with dimensions of no fewer than 299x299 pixels. Images smaller than that are upscaled to this size before being fed to the feature extractor, which may result in poor accuracy.

Typically, scene print works best if the source of your training data matches the source of the images you want to classify. For example, if your app classifies images captured with an iPhone camera, train your model using images captured the same way, if possible.