Amazon Rekognition Moderating Images Documentation

Use this documentation to install and configure Semarchy's plug-in for Amazon Rekognition Moderating Images service.

Overview

Everybody likes using artificial intelligence and machine learning these days, don't they?As part of our strategy to embrace machine learning and AI in more ways, Semarchy is adding AI functionality to xDM as well.

Amazon has been democratizing AI and selling their resulting services. We can benefit from the new Amazon Rekognition service that analyzes images sent to its API and returns information about the image, such as detecting adult content and labels for the content of the images.

This particular article outlines a plug-in enricher Semarchy has built to take advantage of Amazon Rekognition's Moderating Images service that tells you whether there is any adult offensive content in an image.

Plug-ins features

Semarchy's plug-in takes an S3 image URL and passes it into Rekognition which classifies the image according to its Moderating Image service.

See Amazon's documentation for details about this service: Moderating Images documentation

Plug-in Prerequisites

  1. Download the plug-in binary file using this link: Download plug-in.  
  2. Follow these instructions for installing your plug-in: How to install an external plug-in

  3. Prior to using this plug-in, you will need to set up an IAM user with the right IAM roles to access the Amazon Rekognition service. Follow the instructions on Amazon Rekognition's documentation: Authentication and Access Control for Amazon Rekognition.

Plug-In Reference Guide

The following sections detail the parameters, inputs and outputs of the plugin.

Advanced Configuration

The plug-ins enrichers provide options for optimizing and configuring their execution.

Configuration NameDescription
Max RetriesIf the execution of the plug-in fails, it is repeated for this number of times.
Behavior on ErrorIf the execution still fails after the Max Retries have been attempted, the plug-in either skips the current record, skips the entire enrichment task, or stops the whole job, depending on this property.
Thread Pool SizeThis property defines the number of parallel threads used by this plug-in. This option is taken into account if the plug-in used is thread safe and declared as such.

Plug-in Parameters

The following table lists the plug-in parameters.

Parameter NameMandatoryTypeDescriptionSample Value
AWS Access Key ID YesStringProvide the Access Key ID which is generated when a user is created in AWS Identity and Access Management (IAM).AKIAIOSFODNN7EXAMPLE
AWS Secret Access Key YesStringProvide the Secret Access Key which is generated when a user is created in AWS Identity and Access Management (IAM). wJalrXUtnFEMI/K7MDENG/bPxRfiCYEXAMPLEKEY
RegionYesString

Specify in which region you want to use Amazon Rekognition. For example, valid regions include: us-east-1 (US East Northern Virginia), us-west-2 (US West Oregon) and eu-west-1 (EU Ireland) regions. See Amazon docs.

us-east-1

Plug-in Inputs

The following table lists the plug-in inputs.

Parameter NameMandatoryTypeDescriptionSample Value
Image URL YesString

Image stored in S3 to send to Amazon Rekognition for moderation

https://s3.amazonaws.com/semarchy-tutorials/10001-MAIN.jpg
Minimum confidence NoFloatSpecifies the minimum confidence level for the labels to return. Amazon Rekognition doesn't return any labels with a confidence level lower than this specified value. Valid Range: Min 0. Max 100. Plug-in defaults to 0.50

Plug-in Outputs

The following tables list the plug-in outputs.

The output consists of a result status, boolean classification, and confidence scores.


Parameter NameTypeDescriptionSample Value
Is Explicit NudityBooleanBoolean returned. In Oracle, represented by the data type varchar2 and the value 1. See Sample Value.'1'
Explicit Nudity Confidence ScoreFloatA confidence value between 0 and 100 that indicates the percentage confidence that Amazon Rekognition has that the label is correct. See Amazon Rekognition documentation for more details: Moderating Images Documentation.97.06847381591797
Is NudityBooleanBoolean returned. In Oracle, represented by the data type varchar2 and the value 1. See Sample Value.'1'
Nudity Confidence ScoreFloatA confidence value between 0 and 100 that indicates the percentage confidence that Amazon Rekognition has that the label is correct. See Amazon Rekognition documentation for more details: Moderating Images Documentation.97.06847381591797
Is Graphic Male NudityBooleanBoolean returned. In Oracle, represented by the data type varchar2 and the value 1. See Sample Value.'1'
Graphic Male Nudity Confidence ScoreFloatA confidence value between 0 and 100 that indicates the percentage confidence that Amazon Rekognition has that the label is correct. See Amazon Rekognition documentation for more details: Moderating Images Documentation.97.06847381591797
Is Graphic Female NudityBooleanBoolean returned. In Oracle, represented by the data type varchar2 and the value 1. See Sample Value.'1'
Graphic Female Nudity Confidence ScoreFloatA confidence value between 0 and 100 that indicates the percentage confidence that Amazon Rekognition has that the label is correct. See Amazon Rekognition documentation for more details: Moderating Images Documentation.97.06847381591797
Is Sexual ActivityBooleanBoolean returned. In Oracle, represented by the data type varchar2 and the value 1. See Sample Value.'1'
Sexual Activity Confidence ScoreFloatA confidence value between 0 and 100 that indicates the percentage confidence that Amazon Rekognition has that the label is correct. See Amazon Rekognition documentation for more details: Moderating Images Documentation.97.06847381591797
Is Partial NudityBooleanBoolean returned. In Oracle, represented by the data type varchar2 and the value 1. See Sample Value.'1'
Partial Nudity Confidence ScoreFloatA confidence value between 0 and 100 that indicates the percentage confidence that Amazon Rekognition has that the label is correct. See Amazon Rekognition documentation for more details: Moderating Images Documentation.97.06847381591797
Is SuggestiveBooleanBoolean returned. In Oracle, represented by the data type varchar2 and the value 1. See Sample Value.'1'
Suggestive Confidence ScoreFloatA confidence value between 0 and 100 that indicates the percentage confidence that Amazon Rekognition has that the label is correct. See Amazon Rekognition documentation for more details: Moderating Images Documentation.97.06847381591797
Is Female Swimwear Or UnderwearBooleanBoolean returned. In Oracle, represented by the data type varchar2 and the value 1. See Sample Value.'1'
Female Swimwear Or Underwear Confidence ScoreFloatA confidence value between 0 and 100 that indicates the percentage confidence that Amazon Rekognition has that the label is correct. See Amazon Rekognition documentation for more details: Moderating Images Documentation.97.06847381591797
Is Male Swimwear Or UnderwearBooleanBoolean returned. In Oracle, represented by the data type varchar2 and the value 1. See Sample Value.'1'
Male Swimwear Or Underwear Confidence ScoreFloatA confidence value between 0 and 100 that indicates the percentage confidence that Amazon Rekognition has that the label is correct. See Amazon Rekognition documentation for more details: Moderating Images Documentation.97.06847381591797
Is Revealing ClothesBooleanBoolean returned. In Oracle, represented by the data type varchar2 and the value 1. See Sample Value.'1'
Revealing Clothes Confidence ScoreFloatA confidence value between 0 and 100 that indicates the percentage confidence that Amazon Rekognition has that the label is correct. See Amazon Rekognition documentation for more details: Moderating Images Documentation.97.06847381591797
Status StringStatus code to inform whether successful call to Rekognition. If NULL, call failed.Ok

Obtaining Help

If you have trouble with this plug-in, there are many ways to access the Semarchy Technical Support. The easiest is to email our global Technical Support Center (support@semarchy.com). 

Feedback

We welcome your comments and suggestions on the quality and usefulness of this Rekognition plug-in.
If you find any error or have any suggestion for improvement, please mail support@semarchy.com.  

Filter by label

There are no items with the selected labels at this time.