This project demonstrates the process of creating a Veritone face detection engine using the Amazon Rekognition API. For more information about Amazon Rekognition, visit here.
This engine uses node 6.12.2 and requires AWS CLI toolkit.
- Get access to an AWS IAM user with Rekognition access
- Set up aws-cli SDK
- Create a config.json by replacing XXX_PLACEHOLDER variables with corresponding values in the config.json.template.
- Generate a payload.json via VDA
- Build and run the engine:
$ npm install
$ node app.js -config conf/config.json -payload payload.json
- Log into the Veritone Developer Docker registry.
$ docker login docker.veritone.com
- Build the Docker image
$ docker build -t az-rekognition-face-detection .
Once built and tested locally (either as a Docker container or running node app.js
as shown above),
tag the engine and upload the build to the Veritone Docker registry:
$ docker tag az-rekognition-face-detection docker.veritone.com/${YOUR-ACCOUNT}/az-rekognition-face-detection:${CUSTOM-TAG}
$ docker push docker.veritone.com/${YOUR-ACCOUNT}/az-rekognition-face-detection:${CUSTOM-TAG}
After the engine is uploaded it will appear in the the builds table. Press deploy to deploy the engine. Now the engine is now deployed and ready for use.