Facerecognition YOLO

Simple object recognition worker configured to perform face blur on jpg images fed via a RabbitMQ message queue.

The code is part of the final exam of the free anynines Kubernetes Tutorial. Therefore, the code is not meant to be production grade and contains refactoring TODOs that learners may engage with.

The Worker

The worker performs the following steps

• Connect to RabbitMQ
• Retrieve a job (message)
• Retrieve the image
• Start image processing
• Upload the image
• Acknowledge the message (and thus remove it from the queue)

Read the source code for further information.

Building the Image

Example:

docker build -t facerecognition-yolo:dev .

Publishing the Image

Example:

docker tag facerecognition-yolo:dev fischerjulian/facerecognition-yolo:dev
docker push fischerjulian/facerecognition-yolo:dev

As a oneliner:

docker build -t facerecognition-yolo:dev . && docker tag facerecognition-yolo:dev fischerjulian/facerecognition-yolo:dev && docker push fischerjulian/facerecognition-yolo:dev

Running the Image

Example:

docker run --rm -it --name facerecognition-yolo facerecognition-yolo bash

Running the Face Recognition

Inside the container run:

cd python3 yolo_opencv.py --image /tmp/object_recognition/original-image.jpg --config yolov3.cfg --weights yolov3.weights --classes yolov3.txt 

This will produce output file: /tmp/object_recognition/filtered-image.jpg

Downloading the YOLO Weights Definition

wget https://pjreddie.com/media/files/yolov3.weights

Links and Further Reading

1. https://pjreddie.com/darknet/yolo/
2. https://github.com/loretoparisi/docker/tree/master/darknet
3. https://www.arunponnusamy.com/yolo-object-detection-opencv-python.html