Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

A Review of CNN Applications in Smart Agriculture Using Multimodal Data.

Sensors (Basel, Switzerland)·2025
Same author

Convolutional Neural Network for Drowsiness Detection Using EEG Signals.

Sensors (Basel, Switzerland)·2021
See all related articles

Related Experiment Video

Updated: Jul 5, 2025

Deep Neural Networks for Image-Based Dietary Assessment
13:19

Deep Neural Networks for Image-Based Dietary Assessment

Published on: March 13, 2021

9.2K

mid-DeepLabv3+: A Novel Approach for Image Semantic Segmentation Applied to African Food Dietary Assessments.

Thierry Roland Baban A Erep1, Lotfi Chaari1

  • 1Toulouse INP, University of Toulouse, Institut de Recherche en Informatique de Toulouse, 31400 Toulouse, France.

Sensors (Basel, Switzerland)
|January 11, 2024
PubMed
Summary

This study introduces mid-DeepLabv3+, an improved food image segmentation model, and CamerFood10, a new dataset for sub-Saharan African foods. These advancements enhance vision-based dietary assessment systems.

Keywords:
CNNCamerFood10 datasetfood segmentationsemantic segmentation

More Related Videos

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
04:48

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

2.8K
Objectification of Tongue Diagnosis in Traditional Medicine, Data Analysis, and Study Application
05:56

Objectification of Tongue Diagnosis in Traditional Medicine, Data Analysis, and Study Application

Published on: April 14, 2023

2.5K

Related Experiment Videos

Last Updated: Jul 5, 2025

Deep Neural Networks for Image-Based Dietary Assessment
13:19

Deep Neural Networks for Image-Based Dietary Assessment

Published on: March 13, 2021

9.2K
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
04:48

Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

Published on: November 30, 2022

2.8K
Objectification of Tongue Diagnosis in Traditional Medicine, Data Analysis, and Study Application
05:56

Objectification of Tongue Diagnosis in Traditional Medicine, Data Analysis, and Study Application

Published on: April 14, 2023

2.5K

Area of Science:

  • Computer Vision
  • Artificial Intelligence
  • Nutrition Science

Background:

  • Vision-based dietary assessment (VBDA) systems are crucial for nutritional analysis but face challenges in food image segmentation.
  • Existing research primarily focuses on Asian and Western cuisines, neglecting the unique complexities of sub-Saharan African foods, such as high inter-class similarity and mixed-class dishes.

Purpose of the Study:

  • To enhance the food image segmentation stage of VBDA systems for sub-Saharan African cuisine.
  • To introduce an improved segmentation model and a specialized dataset to address existing limitations.

Main Methods:

  • Development of mid-DeepLabv3+, a modified DeepLabv3+ model with a ResNet50 backbone, incorporating an additional decoder layer and SimAM modules.
  • Creation of CamerFood10, the first food image dataset tailored for sub-Saharan African food segmentation, featuring 10 common Cameroonian food items.
  • Evaluation of the mid-DeepLabv3+ model on the CamerFood10 dataset using semantic image segmentation benchmarks.

Main Results:

  • The proposed mid-DeepLabv3+ model achieved a mean Intersection over Union (mIoU) of 65.20% on the CamerFood10 dataset.
  • This represents a significant improvement of +10.74% compared to the standard DeepLabv3+ model with the same ResNet50 backbone.
  • mid-DeepLabv3+ demonstrated superior performance over other benchmark convolutional neural network models for semantic image segmentation in this context.

Conclusions:

  • The mid-DeepLabv3+ model and the CamerFood10 dataset represent significant advancements in food image segmentation for sub-Saharan African diets.
  • These contributions are expected to improve the accuracy and applicability of vision-based dietary assessment systems in diverse populations.
  • Further research can leverage these tools to enhance nutritional monitoring and public health initiatives in the region.