Semi-automated system for predicting calories in photographs of meals

Patrick McAllister, Huiru Zheng, Raymond Bond, Anne Moorhead

Research output: Chapter in Book/Report/Conference proceedingConference contribution

4 Citations (Scopus)

Abstract

Obesity is increasing globally. Obesity brings with it many chronic conditions. There has been increasing research in the use of ICT interventions to combat obesity using food logging and image calorie analysis. These interventions allow users to document their calorie intake to help promote healthy living. However using food logs may lead to inaccurate readings as the user may incorrectly calculate portion size when recording nutritional information. This paper discusses the use of image nutritional analysis techniques to ascertain a more accurate calorie reading from photographs of food items. The methods employed involve determining a ground truth data set by correlating weight of a food item with its area in cm2. This dataset could then be plotted on a regression model and used to determine calorie content of future portions. The proposed system uses a semi-automated approach to allow users to manually draw around the food portion using a polygonal tool. Results show that the application achieved a reasonable accuracy in predicting the calorie content of food item portions with a 11.82% percentage error.
LanguageEnglish
Title of host publicationUnknown Host Publication
Pages1-6
Number of pages6
DOIs
Publication statusPublished - 24 Jun 2015
EventIEEE International Conference on Engineering, Technology and Innovation/ International Technology Management Conference - Belfast
Duration: 24 Jun 2015 → …

Conference

ConferenceIEEE International Conference on Engineering, Technology and Innovation/ International Technology Management Conference
Period24/06/15 → …

Fingerprint

Image analysis

Keywords

  • machine learning
  • machine vision
  • obesity
  • nutrition
  • self-management

Cite this

@inproceedings{a6ad0507cb024c1eb8484157fd1b7959,
title = "Semi-automated system for predicting calories in photographs of meals",
abstract = "Obesity is increasing globally. Obesity brings with it many chronic conditions. There has been increasing research in the use of ICT interventions to combat obesity using food logging and image calorie analysis. These interventions allow users to document their calorie intake to help promote healthy living. However using food logs may lead to inaccurate readings as the user may incorrectly calculate portion size when recording nutritional information. This paper discusses the use of image nutritional analysis techniques to ascertain a more accurate calorie reading from photographs of food items. The methods employed involve determining a ground truth data set by correlating weight of a food item with its area in cm2. This dataset could then be plotted on a regression model and used to determine calorie content of future portions. The proposed system uses a semi-automated approach to allow users to manually draw around the food portion using a polygonal tool. Results show that the application achieved a reasonable accuracy in predicting the calorie content of food item portions with a 11.82{\%} percentage error.",
keywords = "machine learning, machine vision, obesity, nutrition, self-management",
author = "Patrick McAllister and Huiru Zheng and Raymond Bond and Anne Moorhead",
year = "2015",
month = "6",
day = "24",
doi = "10.1109/ICE.2015.7438668",
language = "English",
pages = "1--6",
booktitle = "Unknown Host Publication",

}

McAllister, P, Zheng, H, Bond, R & Moorhead, A 2015, Semi-automated system for predicting calories in photographs of meals. in Unknown Host Publication. pp. 1-6, IEEE International Conference on Engineering, Technology and Innovation/ International Technology Management Conference, 24/06/15. https://doi.org/10.1109/ICE.2015.7438668

Semi-automated system for predicting calories in photographs of meals. / McAllister, Patrick; Zheng, Huiru; Bond, Raymond; Moorhead, Anne.

Unknown Host Publication. 2015. p. 1-6.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Semi-automated system for predicting calories in photographs of meals

AU - McAllister, Patrick

AU - Zheng, Huiru

AU - Bond, Raymond

AU - Moorhead, Anne

PY - 2015/6/24

Y1 - 2015/6/24

N2 - Obesity is increasing globally. Obesity brings with it many chronic conditions. There has been increasing research in the use of ICT interventions to combat obesity using food logging and image calorie analysis. These interventions allow users to document their calorie intake to help promote healthy living. However using food logs may lead to inaccurate readings as the user may incorrectly calculate portion size when recording nutritional information. This paper discusses the use of image nutritional analysis techniques to ascertain a more accurate calorie reading from photographs of food items. The methods employed involve determining a ground truth data set by correlating weight of a food item with its area in cm2. This dataset could then be plotted on a regression model and used to determine calorie content of future portions. The proposed system uses a semi-automated approach to allow users to manually draw around the food portion using a polygonal tool. Results show that the application achieved a reasonable accuracy in predicting the calorie content of food item portions with a 11.82% percentage error.

AB - Obesity is increasing globally. Obesity brings with it many chronic conditions. There has been increasing research in the use of ICT interventions to combat obesity using food logging and image calorie analysis. These interventions allow users to document their calorie intake to help promote healthy living. However using food logs may lead to inaccurate readings as the user may incorrectly calculate portion size when recording nutritional information. This paper discusses the use of image nutritional analysis techniques to ascertain a more accurate calorie reading from photographs of food items. The methods employed involve determining a ground truth data set by correlating weight of a food item with its area in cm2. This dataset could then be plotted on a regression model and used to determine calorie content of future portions. The proposed system uses a semi-automated approach to allow users to manually draw around the food portion using a polygonal tool. Results show that the application achieved a reasonable accuracy in predicting the calorie content of food item portions with a 11.82% percentage error.

KW - machine learning

KW - machine vision

KW - obesity

KW - nutrition

KW - self-management

U2 - 10.1109/ICE.2015.7438668

DO - 10.1109/ICE.2015.7438668

M3 - Conference contribution

SP - 1

EP - 6

BT - Unknown Host Publication

ER -