Real-time and batch prediction service for Openfoodfacts
Project description
Robotoff
Robotoff is a service managing potential Open Food Facts updates (also known as insights). These insights include a growing set of facts, including:
- the product category, weight, brand, packager codes and expiration date
- some of its labels
- abusive pictures (selfies)
- rotated pictures
- ingredient spellchecking
Robotoff provides an API to:
- import a batch of insights in JSONL format
- Fetch insights
- Annotate an insight (accept or reject) and send the update to Openfoodfacts if the insight was accepted
Once generated, the insights can be applied automatically, or after a manual validation if needs be. A scheduler takes care of regularly marking insights for automatic annotation and for sending the update to Openfoodfacts.
The API documentation describes the API endpoints.
For a quickstart of Robotoff as a library, go to the Quickstart.
Installation
Robotoff is made of an API web server, a scheduler, a pool of asynchronous workers and a an Elasticsearch server.
All these services are available as docker images. A docker-compose.yml file is used for service orchestration.
To start all services, simply run:
$ docker-compose up -d
Licence
Robotoff is licenced under the AGPLv3.
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
