The CoreML Player open source project is more of a feasibility study than a productive run-of-the-mill tool, but an exciting software project nonetheless that allows easy access and use of Apple’s Core ML models.

Machine learning with Core ML models

The multimedia player can display and apply photos and videos to compatible Core ML models. Core ML itself is a Machine Learning framework from Apple, based on which statistical models can be created, which then take on tasks such as object recognition or color evaluation.

The CoreML Player can apply the loaded CoreML models to image files and running video content in real time.

The open source project allows you to load your own core ML models, as well as those provided by Apple here. In the basic settings, with a mouse click, the model “YOLOv3-Tiny” can be loaded, which can identify 80 different object classes and then locate them in images and video frames.

All models can be loaded

Other models recognize depth information, can show joint positions for recognized people or recognize handwritten notes.

The CoreML Player must be launched with a right mouse button when first opened and can then be fed with images and videos, over which the Core ML models are applied.

The Mac application is 40 megabytes in size and requires at least macOS 13.0 to be used. The application is completely local and is set to read only the files you choose. The CoreML Player does not have write permissions on your system.