Embedded Files
Back End
Back End
Implementation and Model
Implementation and Model
The data was gathered from users using a Google form. In addition, the data was also added by hand based on drinks found online, in YouTube videos of the "Best Starbucks Drinks on the Secret Menu", and made up by fanatics close to the group.
This data was then painstakingly lablled by the team in LabelStudio and downloaded as a .json file and passed into the spaCy NER model trained in a python Jupyter Notebook.
We needed a way to attach our NLP model to our front-end application. The application would end up being a Flutter application so it would be able to run on both iOS and Android. To do this, we uploaded our model to HuggingFace. On HuggingFace we would be able to access our model as a simple CURL Inference API call. We are currently working with HuggingFace to get a DOI for our model and plan to continue working on this project following the end of the Fall '22 Semester.
Training Data!
Training Data!
We used this to collect preliminary training data
Front End
Front End
Flutter App
Flutter App
Our flutter application uses a two page interface. It is cross-functional between ios and Android as well as hot reload that allows for rapid iteration. This allows us to quick iteration based our stakeholders' feedback.
UI Design: Accessible. Efficient. Coffee.
UI Design: Accessible. Efficient. Coffee.
We designed a UI style guide to ensure consistency across COFF-E's interface, website, and product deliverables. We also crafted a complete icon set to visually represent each ingredient on the menu, expanding universal understandings independent of language labels.
Page updated
Report abuse