AWS Rekognition Serverless

This lambda function will use Rekognition to detect a specific person on thousands of images.

Due to the serverless approach thousands of images can be processed at the same time.

We will create a lambda function / SQS Queue / DynamoDB and IAM Roles.

Prerequisites

• AWS Account
• A S3 bucket containing all your images
• A S3 bucket (can be the same) that contains an image of the target person.

Deploy function:

First deploy the Serverless stack into your AWS account:

export INPUT_BUCKET='your-bucket-name'
serverless deploy -v

Trigger Lambda

Put this item into the SQS queue to trigger the lambda function:

{"img_url": "images/image1.jpg", "target_img_url": "target.jpg"}

You can already use the fill_queue.py script to automatically add all images from a .txt file into that queue.

The results will directly be stored into a Dynamo-DB.

Remove stack

serverless remove