In this project I implemented Automatic face recognition service using kubernetes.
-
rest - The REST frontend will accept images or paths to images for analysis and handle queries concerning specific images. The REST worker will queue tasks to workers using
rabbitmq
messages. -
worker - Worker nodes will receive work requests to analyze images. If those images contain face data that can be scanned, the information is entered into the REDIS database.
-
rabbitmq - The RabbitMQ deployment and service that acts as the rabbit-mq broker.
-
redis - The Redis deployment and service to provide the redis database server.
More details about each of the above can be known through the "README.md" files provided in each of the folders.
The worker node will use open source face recognition software software.