Salesforce Einstein - Create and Train the Dataset - Trailhead Challenge

Question

Trailhead consists of creating a Dataset with the given URL(https://developer.salesforce.com/files/Cats.zip)

But how can we create our own Dataset, if we have all the images in Local system.

Answer 1

The trailhead module used a service class called EinsteinVision_PredictionService to get the data from the zip and then load it into the Einstein REST API..

public Einstein_Dataset createDatasetFromUrlAsync(String url) {
    System.debug('Starting createDatasetFromUrlAsync call from URL ' + url + ' for type ' + typeString + '.');
    Einstein_HttpBodyPartDatasetUrl parts = new Einstein_HttpBodyPartDatasetUrl(url, typeString);
    Einstein_HttpClient client = new Einstein_HttpClient(this, '/datasets/upload', parts.build());
    client.isPost(true);
    System.debug('Target URL is ' + client.getUrl());
    client.execute();
    while(isExecuting()) {
        System.debug('Status is: ' + isExecuting());
    }
    System.debug('Call createDatasetFromUrlAsync has been executed.');
    if (!client.isError()) {
        Einstein_Dataset dataset = (Einstein_Dataset)client.getParser().readValueAs(Einstein_Dataset.class);
        System.debug('New dataset with id ' + dataset.id + ' has been created.');
        return dataset;
    }
    handleError(client.getStatusCode(), client.getResponseError());
    return null;
}

All the source code is here:

• salesforce-einstein-platform-apex

You can follow the Apex code example to understand how they have done it.

The Einstein REST API documentation will also help:

• Introduction to Salesforce Einstein Vision
• Create a Dataset From a Zip File Synchronously