Configure Machine Learning Model

Loads a Core ML (Machine Learning) model and prepares it for use.

Options 

  • Name identifies the model on which the ComputeModel function will operate. See ComputeModel function.
  • Operation specifies how the model is used:
    • Vision assumes that the input is a single image, and the output is an array of classifications. A single classification contains a name and a floating-point confidence level.
    • General takes one or more scalar input values and returns one finite ordered list of results.
    • Unload unloads a model and frees the resources the model consumed. If this operation is used, Name identifies the model to unload.
  • From specifies a container field with the model’s data file (required for all operations except Unload).

Compatibility 

Product Supported
FileMaker Pro Partial
FileMaker Go Yes
FileMaker WebDirect No
FileMaker Server Partial
FileMaker Cloud No
FileMaker Data API No
Custom Web Publishing No

Originated in version 

19.0

Description 

Use this script step to load a Core ML model so that the ComputeModel function can provide the model with its input data and format the output for analysis or display.

The following table explains the Core ML parameter values supported by this script step.

Parameter

Data type in FileMaker Pro

text

Text

int64

Number

double

Number

image

A container field with a bitmapped image type supported by the Apple Core Graphics API

Notes 

  • This script step is supported only on iOS, iPadOS, and macOS.
  • Name can be a calculation.
  • FileMaker products don’t provide the ability to train models.

Example 1 

Loads a vision model from the specified container field.

Copy
Configure Machine Learning Model [ Name: visionModel ; Operation: Vision ; From: ModelContainerField ]

Example 2 

Unloads a model from memory.

Copy
Configure Machine Learning Model [ Name: visionModel ; Operation: Unload ]