AWS Image Recognition
Detects real world objects in images and adds these labels to the repository document on the field simflofy_ai_labels using the AWS Rekognition system.
For your reference, these are ‘MinConfidence’ and ‘MaxLabels’ here. This page is what this task uses.
Configuration- Minimum Threshold: The minimum confidence threshold for labels to return following label detection. Labels with a confidence level lower than this will not be returned.
- Max Number of labels: The maximum number of recognised labels to be returned, by highest confidence.
- Auth conn: Your Amazon AWS credentials
- Max number of labels: The number of labels you want to return
Google Vision Image Labels
Detects real world objects in images and adds these labels to the repository document on the field simflofy_ai_labels using Google Vision.
- Max Number of labels: The maximum number of recognised labels to be returned
- Authentication connection: Your authentication connection for Google. You can find it in the url while edit or view page for the connection
Watson Image Analysis
Uses IBM Watson Image Analysis to analyse an image, adding its response to a specified field.
- The IAM API: Key for IBM Cloud. See instructions here on how to get it.
- Collection ID: If left blank, 3Sixty will create one
- Collection Name: Name of the collection to create. If left blank, will be new-training-collection with the date appended.
- Training Mode: Images will be used in conjunction with the training json to train your analysable. In this mode, no analysis will be performed.
- The training json: Required for training mode. The format is {“dog”:[25,35,105,215]}. Array order is top, left, width, height with object type as the key.
- Field that contains the image urls: If left blank the file content will be used.