The FoodEx2 Smart Coding Application is designed and developed internally in the European Food Safety Authority (DATA and AMU units). This project aims to simplify the FoodEx2 coding process of food starting from a given free text description. More specifically this is possible thanks to the use of FoodEx2 dedicated text classification (single and multi-label) models trained on historycal data.
Find below all the steps needed to run the project.
Clone the project locally in your workspace using the following command:
git clone https://github.com/openefsa/foodex2-sca-back-end.git
Move inside the project's folder just cloned and install all the required dependencies (listed in the requirements.txt file). Create a python virutal environment using the command:
python -m venv venv
Make sure that the virtual environment it is enabled in your terminal or IDE (check here for additional information) and hence, run the following command in order to install all the dependencies using the dedicated python pip package installer:
pip install -r requirements.txt
The file /api/private.py contains services for handling feedback that can be transmitted directly by the user. Feedback must be sent, via a dedicated api (see next section), to the Azure Storage. Once you have successfully created an account and a table (to collect feedback), set the following credentials in the file:
- ACCOUNT_KEY
- ACCOUNT_KEY
- TABLE_NAME
- SECRET CODE
Note that it is also possible to use the application without the feedback engine and therefore Azure Table Storage is not required. To do this, just comment out all the code inside the file.
The file /api/utils.py contains services for handling the translation of non-English food description. The translation is made using Microsoft Azure Cognitive Services Translator. For being able to use the translation function fill in the following credentials:
- TRANSLATOR_SUBSCRIPTION_KEY
- TRANSLATOR_RESOURCE_LOCATION
Note that it is also possible to use the application without the translation function. To do this, just comment out all the code inside the file.
The FoodEx2 SCA back-end project makes use of various text classification models Spacy. For more information on what data and how these models were trained check the wiki.
Note that the project makes use of Kubernates which automatically downloads and extract all models into the dedicated folder. For testing purposes, without using Kubernates, just extract the models within the models folder.
Several APIs are available on the FoodEx2 Smart Coding Application (check the wiki for additional information). To expose them, you need to serve the project using Flask. Run the following command from the project folder:
python app.py
The following section describes how to deploy the FoodEx2 Smart Coding Application's back-end locally.
Install Docker and configure it on your local computer. From the main folder of the foodex2-sca-backend project run the following command:
docker build -t name:tag
This command will use the DOCKERFILE, present in the main folder, for creating the docker image. Check if the docker image is present by launching the following command:
docker images
After making sure that the docker image has been correcly created, run it with the following command:
docker run name:tag
Deploy the docker image created on kubernates by using the files present under the "/manifest" folder by running the following command:
kubectl deploy -f create ./manifest/deployment.yml