Before any training or classification can occur, a Model must be created. A Model is an object which contains all the knowledge from the training that will allow it to recognise images.

• There can be multiple Models within a site, each working independently with their own training.
• You need to follow the Model creating process even if there will only be one Model within the site

Create a Model Table

Start by creating a Table to represent the Models.

This should have the following fields:

• RecordID (type RecordID)
• ModelID (type ModelID)
• Model Name

Note: Take care not to confuse the RecordID, which is the unique ID for the Record in the Table, with the ModelID, which will represent the ML Model object.

Create a Model

Configure an ML - Create Model event action

In the Model table, Record Change Event, add an action "ML - Create Model"

Make sure this action is only executed once for each record: so typically you would add criteria to make it only execute if the record is New.

In the Configuration tab:

Input

Model Name

Set this to the field in the table where the Model Name has been entered.

User ID

Models can be associated with particular users. 

Return

Model Id

When the action has completed, it will generate a value for the ModelID and place that in the Model ID field. This will be used in subsequent actions to identify the model to work with

Diagnostics

This section gives the standard event action diagnostic feedback.

Completed OK

Store in a Checkbox field

Feedback

Store in a Text field