Date of Award

6-13-2018

Document Type

Thesis

Department

Computer Engineering

First Advisor

Moe Amouzgar

Abstract

Many people desire to be informed about the nutritional specifics of the food they consume. Current popular dietary tracking methods are too slow and tedious for a lot of consumers due to requiring manual data entry for everything eaten. We propose a system that will take advantage of image recognition and the internal camera of Android phones to identify food based off of a picture of a user’s plate. Over the course the last year, we trained an object detection model with images of different types of food, built a mobile application around it, and tested their integration and performance. We believe that our program meets the requirements we set out for it at its conception and delivers a simple, fast, and efficient way of tracking one’s diet.

Download

Share

COinS