thumbor_rekognition

Enable thumbor to use AWS rekognition to run face detection

thumbor is a smart imaging service. It enables on-demand crop, resizing and flipping of images. It also features a smart detection of important points in the image for better cropping and resizing, using opencv face detection algorithms.

AWS Rekognition can quickly detect faces in image using sophisticated deep learning-based visual search and image classification.

thumbor_rekognition helps to use AWS Rekognition service to do face detection in thumbor easily.

Installation

pip install thumbor_rekognition

Enable detector

Add thumbor_rekognition to your thumbor.conf. Do not include thumbor's out-of-box face detector.

DETECTORS = [
    # Do not use out-of-box face detector
    # 'thumbor.detectors.face_detector',

    # Include thumbor_rekognition
    'thumbor_rekognition',
    'thumbor.detectors.feature_detector',
]

The above configuration tells thumbor that it should run both the facial detection and the feature detection. These are mutually exclusive, meaning that if a face is detected, the feature detector won’t be run.

AWS Rekognition

Authentication

Authentication is handled by botocore, see Boto3 documentation.

Region

To set the region to use Rekognition service, add REKOGNITION_REGION to your thumbor.conf. Default is us-east-1.

# AWS region for the Rekognition service
REKOGNITION_REGION = 'us-west-2'

Using it

After installing thumbor_rekognition, you can test it by using the following URL.

http://<thumbor>:<port>/unsafe/300x300/smart/my.domain.com/picture.png

If you need to see what thumbor is seeing in your image, prepend /debug to all the parameters.

http://<thumbor>:<port>/unsafe/debug/300x300/smart/my.domain.com/picture.png