Towards Personalised Training of Machine Learning Algorithms for Food Image Classification Using a Smartphone Camera

Patrick McAllister, Huiru Zheng, Raymond Bond, Anne Moorhead

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

3 Citations (Scopus)

Abstract

This work is related to the development of a personalised machine learning algorithm that is able to classify food images for food logging. The algorithm would be personalised as it would allow users to decided what food items the model will be able to classify. This novel concept introduces the idea of promoting dietary monitoring through classifying food images for food logging by personalising a machine learning algorithm. The food image classification algorithm will be trained based on specific types of foods decided by the user (most popular foods, food types e.g. vegetarian). This would mean that the classification algorithm would not have to be trained using a wide variety of foods which may lead to low accuracy rate but only a small number of foods chosen by the user. To test the concept, a range of experiments were completed using 30 different food types. Each food category contained 100 images. To train a classification algorithm, features were extracted from each food type, features such as SURF, LAB colour features, SFTA, and Local Binary Patterns were used. A number of classification algorithms were used in these experiments; Nave Bayes, SMO, Neural Networks, and Random Forest. The highest accuracy achieved in this work was 69.43 % accuracy using Bag-of-Features (BoF) Colour, BoF-SURF, SFTA, and LBP using a Neural Network.
LanguageEnglish
Title of host publicationUnknown Host Publication
Pages178-190
Number of pages12
Volume10069
DOIs
Publication statusE-pub ahead of print - 2 Nov 2016
Event10th International Conference on Ubiquitous Computing and Ambient ‪Intelligence - Spain
Duration: 2 Nov 2016 → …

Conference

Conference10th International Conference on Ubiquitous Computing and Ambient ‪Intelligence
Period2/11/16 → …

Fingerprint

Image classification
Smartphones
Learning algorithms
Learning systems
Cameras
Color
Neural networks

Keywords

  • Machine learning
  • machine vision
  • image processing
  • health informatics
  • obesity
  • smart phones
  • mobile computing

Cite this

@inproceedings{e447209d6fb0419196d30a77dcbc9d35,
title = "Towards Personalised Training of Machine Learning Algorithms for Food Image Classification Using a Smartphone Camera",
abstract = "This work is related to the development of a personalised machine learning algorithm that is able to classify food images for food logging. The algorithm would be personalised as it would allow users to decided what food items the model will be able to classify. This novel concept introduces the idea of promoting dietary monitoring through classifying food images for food logging by personalising a machine learning algorithm. The food image classification algorithm will be trained based on specific types of foods decided by the user (most popular foods, food types e.g. vegetarian). This would mean that the classification algorithm would not have to be trained using a wide variety of foods which may lead to low accuracy rate but only a small number of foods chosen by the user. To test the concept, a range of experiments were completed using 30 different food types. Each food category contained 100 images. To train a classification algorithm, features were extracted from each food type, features such as SURF, LAB colour features, SFTA, and Local Binary Patterns were used. A number of classification algorithms were used in these experiments; Nave Bayes, SMO, Neural Networks, and Random Forest. The highest accuracy achieved in this work was 69.43 {\%} accuracy using Bag-of-Features (BoF) Colour, BoF-SURF, SFTA, and LBP using a Neural Network.",
keywords = "Machine learning, machine vision, image processing, health informatics, obesity, smart phones, mobile computing",
author = "Patrick McAllister and Huiru Zheng and Raymond Bond and Anne Moorhead",
year = "2016",
month = "11",
day = "2",
doi = "10.1007/978-3-319-48746-5_18",
language = "English",
isbn = "978-3-319-48745-8",
volume = "10069",
pages = "178--190",
booktitle = "Unknown Host Publication",

}

McAllister, P, Zheng, H, Bond, R & Moorhead, A 2016, Towards Personalised Training of Machine Learning Algorithms for Food Image Classification Using a Smartphone Camera. in Unknown Host Publication. vol. 10069, pp. 178-190, 10th International Conference on Ubiquitous Computing and Ambient ‪Intelligence, 2/11/16. https://doi.org/10.1007/978-3-319-48746-5_18

Towards Personalised Training of Machine Learning Algorithms for Food Image Classification Using a Smartphone Camera. / McAllister, Patrick; Zheng, Huiru; Bond, Raymond; Moorhead, Anne.

Unknown Host Publication. Vol. 10069 2016. p. 178-190.

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

TY - GEN

T1 - Towards Personalised Training of Machine Learning Algorithms for Food Image Classification Using a Smartphone Camera

AU - McAllister, Patrick

AU - Zheng, Huiru

AU - Bond, Raymond

AU - Moorhead, Anne

PY - 2016/11/2

Y1 - 2016/11/2

N2 - This work is related to the development of a personalised machine learning algorithm that is able to classify food images for food logging. The algorithm would be personalised as it would allow users to decided what food items the model will be able to classify. This novel concept introduces the idea of promoting dietary monitoring through classifying food images for food logging by personalising a machine learning algorithm. The food image classification algorithm will be trained based on specific types of foods decided by the user (most popular foods, food types e.g. vegetarian). This would mean that the classification algorithm would not have to be trained using a wide variety of foods which may lead to low accuracy rate but only a small number of foods chosen by the user. To test the concept, a range of experiments were completed using 30 different food types. Each food category contained 100 images. To train a classification algorithm, features were extracted from each food type, features such as SURF, LAB colour features, SFTA, and Local Binary Patterns were used. A number of classification algorithms were used in these experiments; Nave Bayes, SMO, Neural Networks, and Random Forest. The highest accuracy achieved in this work was 69.43 % accuracy using Bag-of-Features (BoF) Colour, BoF-SURF, SFTA, and LBP using a Neural Network.

AB - This work is related to the development of a personalised machine learning algorithm that is able to classify food images for food logging. The algorithm would be personalised as it would allow users to decided what food items the model will be able to classify. This novel concept introduces the idea of promoting dietary monitoring through classifying food images for food logging by personalising a machine learning algorithm. The food image classification algorithm will be trained based on specific types of foods decided by the user (most popular foods, food types e.g. vegetarian). This would mean that the classification algorithm would not have to be trained using a wide variety of foods which may lead to low accuracy rate but only a small number of foods chosen by the user. To test the concept, a range of experiments were completed using 30 different food types. Each food category contained 100 images. To train a classification algorithm, features were extracted from each food type, features such as SURF, LAB colour features, SFTA, and Local Binary Patterns were used. A number of classification algorithms were used in these experiments; Nave Bayes, SMO, Neural Networks, and Random Forest. The highest accuracy achieved in this work was 69.43 % accuracy using Bag-of-Features (BoF) Colour, BoF-SURF, SFTA, and LBP using a Neural Network.

KW - Machine learning

KW - machine vision

KW - image processing

KW - health informatics

KW - obesity

KW - smart phones

KW - mobile computing

U2 - 10.1007/978-3-319-48746-5_18

DO - 10.1007/978-3-319-48746-5_18

M3 - Conference contribution

SN - 978-3-319-48745-8

VL - 10069

SP - 178

EP - 190

BT - Unknown Host Publication

ER -