Fruit of an acquisition: Apple AI software goes open
Turi Create lands on GitHub
Apple's joined other juggernauts of the tech sector by releasing an open source AI framework.
As the GitHub description explains, it targets app developers that want custom machine learning models but don't have the expertise to “add recommendations, object detection, image classification, image similarity or activity classification” to their apps.
Completed models are exported to Core ML for use in “iOS, macOS, watchOS, and tvOS apps”.
Other details noted at the repo include a focus on tasks rather than algorithms; built-in streaming visualisation for data exploration; support for text, images, audio, video, and sensor data; and it can “work with large datasets on a single machine”.
To illustrated ease of use, the repo shows the following code for image recognition:
import turicreate as tc # Load data data = tc.SFrame('photoLabel.sframe') # Create a model model = tc.image_classifier.create(data, target='photoLabel') # Make predictions predictions = model.predict(data) # Export to Core ML model.export_coreml('MyClassifier.mlmodel')
Models supported in Turi Create include classifiers, regression, graph analytics, clustering, nearest neighbour, and topic models.
Developers can run Turi Create on Mac OS 10.12 or higher, Linux with glibc 2.12 or higher, or Windows 10 through the Windows Subsystem for Linux. ®