Recipes training data samples upload
Automatic recipe processing is a microservice pipeline for processing and extracting recipes obtained from images or websites. For the extraction of relevant recipe information (recipe name, ingredients, description...) Named Entity Recognition (NER) is used.
python -m venv venv/
source venv/bin/activate
pip install .
Start the FastAPI web server in development mode: uvicorn main:app
[
{
"id": "image1.png",
"text": "Käsekuchen\n\n3 Portionen...",
"ocr": <Text object>
},
...
]
POST requests for uploading files can be done via the following:
curl --location --request POST 'http://localhost:8080/uploadfiles' --form 'files=@"file1.txt"' --form 'files=@"file2.txt' ...
Upload files via the UI at http://localhost:8080/upload
.
docker build -t arp-demo .
docker run --rm -v $PWD/data:/data -p <localport>:80 arp-demo