01· the project
> A project for Sunny Side Up Studios.
One photo, and the dish has a name.
Kiweat recognises dishes from a photograph. Take a picture of a plate and it tells you what is on it, out of a catalogue of five hundred dishes. We built the recognition with TensorFlow and the Django and MongoDB back end behind it.
The catalogue is a list of 500 dishes, and the recognition model that tells them apart runs in TensorFlow. Behind it sits a Python back end in Django, with MongoDB for the data, in three parts: the app, the model and the database. We built it in 2016.
02· in numbers
500> dishes in the catalogue
3> parts: app, model, database
2016> TensorFlow and Django
03· how it works
01
Photograph
Take a picture of a plate.
02
Recognise
A TensorFlow model reads the image.
03
Match
The dish is named from a catalogue of 500.
04
Answer
A Django back end with MongoDB returns the result.
04· credits
for sunny side up studios. AI. 2016
Working on something like this? Tell us what you need.
05· next project
> cd /works/espadaysantacruz-web
Espadaysantacruz> Web
