README

Akeneo Rekognition Bundle - C&M

Akeneo bundle to enrich products with objects and texts detected in images.

Akeneo Rekognition Bundle allows to retrieve objects and texts detected with AWS Rekognition (using rekognition-php) from a product model image and to store them into this product model.

Versions

An AWS account is also required as AWS Rekognition will be used.

Installation

Download the Bundle

$ composer require clickandmortar/akeneo-rekognition-bundle

Enable the Bundle

Enable the bundle by adding it to the list of registered bundles in the config/bundles.php file of your project:

<?php

return [
    ...
    ClickAndMortar\AkeneoRekognitionBundle\ClickAndMortarAkeneoRekognitionBundle::class => ['all' => true]
    ...
];

Configuration

Configure credentials

Before using Akeneo Rekognition Bundle, set credentials to make requests to Amazon Web Services.

In config/services/services.yml:

parameters:
    ...
    aws_access_key_id: <your_access_key_id>
    aws_secret_access_key: <your_secret_access_key>
    aws_rekognition_minimum_confidence: 75
    ...

Import attributes

Import new attributes to store data from Rekognition:

php bin/console akeneo:batch:job -c "{\"filePath\":\"vendor/clickandmortar/akeneo-rekognition-bundle/Resources/fixtures/attributes.csv\"}" <your_attribute_import_job_code>

Add new attributes to family

Add new attributes to family

Edit a family variant (if necessary)

Edit a family variant

Create job

php bin/console akeneo:batch:create-job internal add_rekognition_data mass_edit add_rekognition_data '{}' 'Add Rekognition Data'

Usage

Run job

The following line will process all products and models and add data from Rekognition if attributes are editable from family.

php bin/console akeneo:batch:job add_rekognition_data

Mass edit

From products / models list:

• Check the ones that need to be processed.
• Click "Mass edit".
• Click "Add Rekognition Data".
• Click "Next", "Next", then "Confirm".
• Check on dashboard that operation has status Completed.

Open products / models that were previously checked. They now have attributes filled with Rekognition data (only if attributes are editable)

Roadmap

• Handle locale as currently labels are in english
• Add fields to store more information provided by Rekognition
• Add composer post install to avoid to play some configuration commands manually
• Find a way to use environment variables with php-fpm (credentials AWS) for docker