How to Use AWS Rekognition ( Image Recognition by AWS )

Learn All About AWS Rekognition

What is AWS Rekognition ?

Amazon Rekognition AWS is an image analysis and recognition software which was built on the top of deep learning , deep learning is an AI system which is used to analyse and recognize images, with AWS Rekognition you can use all the image analysis features without the need to build or learn any of the deep learning technologies, AWS Rekognition was released on November 2016 as the first cloud image analysis service by Amazon.

AWS Rekognition will make image processing easier for developers and anyone who wants to pass a picture to an API and get a JSON file showing information about the picture.

Some of AWS Rekognition Features:
  • Face analysis.
  • Estimate age range for people.
  • Face comparison.
  • Objects detection and analysis (Tree, Grass, Moon , Light...etc).
  • Connect it with CloudWatch. (Get number of detected faces, response time....etc)
Try AWS Rekognition
  • Login to AWS Console.
  • Search for Rekognition.
  • Click  'Try Demo' button.
  • Now you can upload a picture and experience how AWS Rekognition will recognize it and tell it's features.

AWS Rekognition Tutorial


How to Use AWS Rekognition ?
AWS Rekognition SDK Example
  • Create a bucket and place a picture in it.
  • Find name of the bucket and region.
  • Run following command from the EC2 and you will get an output with the recognized objects.

aws rekognition detect-labels --image "S3Object={Bucket=name-of-bucket,Name=test.jpg}" --region us-west-2

Output example:

 "Labels": [
 "Confidence": 99.31684875488281,
 "Name": "Human"
 "Confidence": 99.31802368164062,
 "Name": "People"
 "Confidence": 99.31802368164062,
 "Name": "Person"
 "Confidence": 97.01766204833984,
 "Name": "Plant"
 "Confidence": 97.01766204833984,
 "Name": "Potted Plant"
 "Confidence": 71.12362670898438,
 "Name": "Leisure Activities"
AWS Rekognition Cost

You can use up to 5000 pictures free a month as a part of the free tier from Amazon which will be enough for testing and 1000 face recognition metadata after that it will be $1 per 1000 calls.

More resources:

Subscribe to
for video tutorials updates