filiphagan / learning-coco-processing-container

COCO data processing with Docker

Description

Learning task: Data processing container transforming COCO-formatted data to CSV. In this example the script expects the URL to zipped archive including person_keypoints_val2017.json

Requirements: docker

Contents

• main.py
• Dockerfile
• requirements.txt
• README.md

Execution

1. Build the docker image

docker build -t mlops-coco <path/to/Dockerfile>

2. Run the container

docker run --name my-coco -t -d mlops-coco

3. Run processing script

Parameters:
--url (str): url to zipped COCO-formatted data package including person_keypoints_val2017.json
--output (str): path to output csv file

docker exec my-coco /code/main.py --url <https://url/to/filename.zip> --output <path/filename.csv>

# Example:
docker exec my-coco python /code/main.py --url "http://images.cocodataset.org/annotations/annotations_trainval2017.zip" --output "/output/out.csv"

Check if the file exists in output directory

docker exec my-coco sh -c "cd /output; ls -l"