Rekognizing patterns

Introduction to AWS Boto in Python

Maksim Pecherskiy

Instructor

Rekognition

So what is Rekognition anyway?

Detecting Objects in an image

Detect objects in image

Extracting Text from Images

Extract text from image

Why not build our own?

Use Rekognition if:

Quick but good
Keep code simple
Recognize many things

Build a model if:

Custom requirements
Security implications
Large volumes

Rekognition screenshot

I am not a computer vision expert

Not a Computer Vision Expert

Aws Docs

BOTO DOCUMENTATION

Upload an image to S3

Initialize S3 Client

s3 = boto3.client(
  's3', region_name='us-east-1', 
  aws_access_key_id=AWS_KEY_ID, aws_secret_access_key=AWS_SECRET
)

Upload a file

s3.upload_file(
  Filename='report.jpg', Key='report.jpg', 
  Bucket='datacamp-img')

Object detection

Initiate the client

rekog = boto3.client(
  'rekognition',
  region_name='us-east-1', 
  aws_access_key_id=AWS_KEY_ID, 
  aws_secret_access_key=AWS_SECRET)

Initiate the Rekognition client

Object detection

Detect!

response = rekog.detect_labels(
  Image={'S3Object': {
          'Bucket': 'datacamp-img', 
          'Name': 'report.jpg'
        },

    MaxLabels=10,
    MinConfidence=95
)

Object detection

Detect labels response

Detect Labels Response

Text detection

Perform detection

response = rekog.detect_text(
  Image={'S3Object': 
          {
            'Bucket': 'datacamp-img', 
            'Name': 'report.jpg'
          }
        }
)

Text detection response

Summary

Detect objects in an image
Count instances
Learn a new AWS Service

Rekognition

.detect_labels()

Recognize text in an image
Word vs line detections
When to build our own

Text detection

.detect_text()

Let's do some computer vision!

Introduction to AWS Boto in Python