webleads-tracker

Food tray Recognition - Devoteam G Cloud - Revendeur Google Apps et intégrateur des solutions Google Entreprise
Food tray Recognition

Devoteam G Cloud > Actualités > Actu Devoteam G Cloud > Food tray Recognition

Food tray Recognition

Publié le : 26 octobre 2018

 

Abstract

With almost one million meals cooked every day for companies, our customer is one of the french leaders in company restaurants. Devoted to innovation, this company improves continuously its offer and services. They have a dedicated team in charge of experiment new use cases for restaurants : ergonomics, flows of work and guests, decoration and arrangement for an optimal development of the services.

Customer Challenges

One of the biggest challenge to improve the user experience at their restaurants is to fast the waiting time at the checkout cashier. On traditional way of working, the cashier needs to type every dish on the tray on his/her machine. The main stake was : “how to automate the collection process to reduce queues ?”

Based on Google Machine Learning, Devoteam G Cloud supported our customer in the dissociation, recognition and automatic classification of the elements of each tray.

Benefits

The custom model we trained with tensorflow gave excellent results for recognition and automatic billing. It shorted the waiting time and increased the productivity.

Implemented GCP Service

  • Prototyping on Colab notebook
  • Tensorflow Object Detection
  • Application App Engine
  • Vision API

Vision API 

Capture d’écran 2018-10-26 à 14.35.08

 

 

 

 

 

 

 AUTO ML Vision

Capture d’écran 2018-10-26 à 14.35.14

 

 

 

 

Objet detection 

Capture d’écran 2018-10-26 à 14.35.20

 

 

 

 

Instance segmentation

Capture d’écran 2018-10-26 à 14.35.25