{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,8,25]],"date-time":"2026-08-25T06:23:50Z","timestamp":1787639030181,"version":"build-2736575974"},"reference-count":47,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,8,25]],"date-time":"2026-08-25T00:00:00Z","timestamp":1787616000000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Public Health"],"abstract":"<jats:sec>\n                    <jats:title>Introduction<\/jats:title>\n                    <jats:p>The ThaiSook mHealth application was designed to promote a healthy lifestyle and mitigate risk factors associated with noncommunicable diseases; one of the administrative tasks involved is the manual flagging and reporting of incorrect entries related to fruit and vegetable intake. However, the process can be time-consuming and prone to errors. Automating food categorisation can provide immediate insights into dietary choices and encourage healthier eating habits. Therefore, we aimed to develop a food image classification model that can be used to categorise different types of Southeast Asian fruits and vegetables, for which nutritional data are available from the Thailand Ministry of Public Health, using the ThaiSook mHealth application.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>A two-stage model encompassing binary and multiclass classification stages was developed to classify 46 types of Southeast Asian fruit, vegetables, and vegetable-heavy dishes. We used a two-stage strategy featuring binary and multiclass models; EfficientNetV2S emerged as the optimal choice for the classification model. The binary classification phase eliminated non-fruit and non-vegetable images, whereas the subsequent multiclass classification identified specific fruit or vegetable types. The methodology incorporates semi-supervised learning using Noisy Student training, complemented by techniques such as pseudo-labelling, label smoothing, and temperature scaling. These strategies enhance model performance and robustness and effectively exploit unlabelled data.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>The dataset included the labelled group (5,760 fruit and vegetable images and 5,450 non-fruit or vegetable images) and unlabelled group (29,385 images; contributed by ThaiSook application users). Our system\u2019s comprehensive performance showed that the end-to-end pipeline achieved accuracies of 92.43, 94.56, and 94.91% in the top-1, \u22123, and \u22125 experiments, respectively.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Conclusion<\/jats:title>\n                    <jats:p>We developed a robust image classification model tailored to fruits and vegetables that can be practically applied for dietary assessment using the ThaiSook mHealth application. Our findings demonstrate the resilience and suitability of our model for resource-constrained mobile applications.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Trial registration<\/jats:title>\n                    <jats:p>The clinical trial was registered according to the WHO International Clinical Trials Registry Platform (WHO-ICTRP) in the Thai Clinical Trials Registry (ID TCTR20220611001) on 11 June 2022.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.3389\/fpubh.2026.1875869","type":"journal-article","created":{"date-parts":[[2026,8,25]],"date-time":"2026-08-25T05:34:46Z","timestamp":1787636086000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["A two-stage deep-learning framework for image classification of 46 southeast Asian fruit and vegetable categories"],"prefix":"10.3389","volume":"14","author":[{"given":"Decho","family":"Surangsrirat","sequence":"first","affiliation":[{"name":"Digital Healthcare Platform Innovation Group, National Science and Technology Development Agency","place":["Pathum Thani, Thailand"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Panyawut","family":"Sri-iesaranusorn","sequence":"additional","affiliation":[{"name":"Digital Healthcare Platform Innovation Group, National Science and Technology Development Agency","place":["Pathum Thani, Thailand"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Warisara","family":"Asawaponwiput","sequence":"additional","affiliation":[{"name":"Canterbury Technical Institute","place":["Brisbane, QLD, Australia"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Wassapon","family":"Watanakeesuntorn","sequence":"additional","affiliation":[{"name":"D3 Center, Osaka University","place":["Osaka, Japan"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Polathep","family":"Vichitkunakorn","sequence":"additional","affiliation":[{"name":"Department of Family and Preventive Medicine, Faculty of Medicine, Prince of Songkla University","place":["Songkla, Thailand"]},{"name":"Health Policy Research Center, Faculty of Medicine, Prince of Songkla University","place":["Songkhla, Thailand"]},{"name":"Research and Innovation Center for Well-being and Continuing Care, Prince of Songkla University, Hat Yai","place":["Songkhla, Thailand"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2026,8,25]]},"reference":[{"key":"ref1","doi-asserted-by":"publisher","first-page":"e18907","DOI":"10.2196\/18907","article-title":"Wearable health devices in health care: narrative systematic review","volume":"8","author":"Lu","year":"2020","journal-title":"JMIR Mhealth Uhealth"},{"key":"ref2","year":""},{"key":"ref3","year":""},{"key":"ref4","year":"2020"},{"key":"ref5","author":"Islam","year":"2018"},{"key":"ref6","author":"Yadav","year":"2021"},{"key":"ref7","author":"Ahmed","year":"2021"},{"key":"ref8","author":"Sumanth","year":"2024"},{"key":"ref9","doi-asserted-by":"publisher","first-page":"591","DOI":"10.1016\/j.matpr.2021.06.016","article-title":"An integrated approach using CNN-RNN-LSTM for classification of fruit images","volume":"51","author":"Gill","year":"2022","journal-title":"Mater Today Proc"},{"key":"ref10","author":"G","year":"2026"},{"key":"ref11","doi-asserted-by":"publisher","first-page":"108395","DOI":"10.1016\/j.jfca.2025.108395","article-title":"Fine-grained food image recognition using a convolutional neural network and swin transformer hybrid model","volume":"148","author":"Xiao","year":"2025","journal-title":"J Food Compos Anal"},{"key":"ref12","doi-asserted-by":"publisher","first-page":"1704.04861","DOI":"10.48550\/arXiv.1704.04861","article-title":"MobileNets: efficient convolutional neural networks for mobile vision applications","author":"Howard","year":"2017","journal-title":"arXiv"},{"key":"ref13","doi-asserted-by":"publisher","first-page":"1905.02244","DOI":"10.48550\/arXiv.1905.02244","article-title":"Searching for MobileNetV3 architecture","author":"Howard","year":"2019","journal-title":"arXiv"},{"key":"ref14","doi-asserted-by":"publisher","first-page":"2249","DOI":"10.1016\/j.procs.2021.08.238","article-title":"Mobilenet family tailored for raspberry pi","volume":"192","author":"Glego\u0142a","year":"2021","journal-title":"Procedia Comput Sci"},{"key":"ref15","doi-asserted-by":"publisher","first-page":"140","DOI":"10.3390\/electronics11010140","article-title":"Optimized deep learning algorithms for tomato leaf disease detection with hardware deployment","volume":"11","author":"Tarek","year":"2022","journal-title":"Electronics (Basel)"},{"key":"ref16","first-page":"450","volume-title":"Mobile-hand: Real-time 3d hand shape and pose estimation from color image","author":"Lim","year":"2020"},{"key":"ref17","doi-asserted-by":"publisher","first-page":"970","DOI":"10.1016\/j.amjmed.2015.03.027","article-title":"American Heart Association\u2019s life\u2019s simple 7: avoiding heart failure and preserving cardiac structure and function","volume":"128","author":"Folsom","year":"2015","journal-title":"Am J Med"},{"key":"ref18","doi-asserted-by":"publisher","first-page":"1909","DOI":"10.1161\/STROKEAHA.111.000352","article-title":"Life\u2019s simple 7 and risk of incident stroke: the reasons for geographic and racial differences in stroke (REGARDS) study","volume":"44","author":"Kulshreshtha","year":"2013","journal-title":"Stroke"},{"key":"ref19","year":"2016"},{"key":"ref20","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.1412.6980","article-title":"Adam: a method for stochastic optimization","author":"Kingma","year":"2014","journal-title":"arXiv"},{"key":"ref21","doi-asserted-by":"publisher","DOI":"10.48550\/ARXIV.1905.11946","article-title":"EfficientNet: rethinking model scaling for convolutional neural networks","author":"Tan","year":"2019","journal-title":"arXiv"},{"key":"ref22","first-page":"355","volume-title":"Lecture Notes in Networks and Systems","author":"Shah","year":"2025"},{"key":"ref23","doi-asserted-by":"publisher","first-page":"105326","DOI":"10.1016\/j.compag.2020.105326","article-title":"Automated fruit recognition using EfficientNet and MixNet","volume":"171","author":"Duong","year":"2020","journal-title":"Comput Electron Agric"},{"key":"ref24","doi-asserted-by":"publisher","first-page":"724","DOI":"10.1002\/cyto.a.20430","article-title":"A multi-model approach to simultaneous segmentation and classification of heterogeneous populations of cell nuclei in 3d confocal microscope images","author":"Lin","year":"2007","journal-title":"Cytometry A"},{"key":"ref25","first-page":"83","volume-title":"ISBI 2019 C-NMC Challenge: Classification in Cancer Cell Imaging. Lecture Notes in Bioengineering","author":"Xiao","year":"2019"},{"key":"ref26","first-page":"896","volume-title":"Pseudo-label: The simple and efficient semi-supervised learning method for deep neural networks. IC","author":"Lee","year":"2013"},{"key":"ref27","first-page":"15094","volume-title":"IEEE\/CVF Conference on Computer Vision and Pattern Recognition","author":"Hu","year":"2021"},{"key":"ref28","first-page":"12961","volume-title":"IEEE\/CVF Conference on Computer Vision and Pattern Recognition","author":"Su","year":"2021"},{"key":"ref29","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2206.06359","article-title":"Energymatch: energy-based pseudo-labeling for semi-supervised learning","author":"Yu","year":"2022","journal-title":"arXiv"},{"key":"ref30","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2201.10836","article-title":"PARS: pseudo-label aware robust sample selection for learning with noisy labels","author":"Goel","year":"2022","journal-title":"Clin Orthop Relat Res"},{"key":"ref31","first-page":"2818","volume-title":"IEEE Conference on Computer Vision and Pattern Recognition (CVPR)","author":"Szegedy","year":"2016"},{"key":"ref32","doi-asserted-by":"publisher","first-page":"2009.06432","DOI":"10.48550\/arXiv.2009.06432","article-title":"Adaptive label smoothing","author":"Krothapalli","year":"2020","journal-title":"arXiv"},{"key":"ref33","doi-asserted-by":"publisher","first-page":"437","DOI":"10.1080\/2150704X.2016.1149252","article-title":"A structural similarity-based label-smoothing algorithm for the post-processing of land-cover classification","volume":"7","author":"Lu","year":"2016","journal-title":"Remote Sens Lett"},{"key":"ref34","first-page":"1321","article-title":"International conference on machine learning","volume-title":"On calibration of modern neural networks","author":"Guo","year":"2017"},{"key":"ref35","first-page":"61","volume-title":"Probabilistic Outputs for Support Vector Machines and Comparisons to Regularized Likelihood Methods","author":"Platt","year":"2000"},{"key":"ref36","author":"Ji","year":"2019"},{"key":"ref37","doi-asserted-by":"publisher","first-page":"1810.11586","DOI":"10.48550\/arXiv.1810.11586","article-title":"Attended temperature scaling: a practical approach for calibrating deep neural networks","author":"Mozafari","year":"2018","journal-title":"arXiv"},{"key":"ref38","author":"Xie","year":"2020"},{"key":"ref39","doi-asserted-by":"publisher","first-page":"1929","DOI":"10.5555\/2627435.2670313","article-title":"Dropout: a simple way to prevent neural networks from overfitting","volume":"15","author":"Srivastava","year":"2014","journal-title":"J Mach Learn Res"},{"key":"ref40","author":"Cubuk","year":"2020"},{"key":"ref41","author":"Oltean","year":"2021"},{"key":"ref42","doi-asserted-by":"publisher","first-page":"33269","DOI":"10.1007\/s11042-022-12868-2","article-title":"Fruit recognition from images using deep learning applications","volume":"81","author":"Gill","year":"2022","journal-title":"Multimed Tools Appl"},{"key":"ref43","doi-asserted-by":"publisher","first-page":"416","DOI":"10.1016\/j.ijcce.2024.09.003","article-title":"A multi-fused convolutional neural network model for fruit image classification","volume":"5","author":"Sinha","year":"2024","journal-title":"Int J Cogn Comput Eng"},{"key":"ref44","author":"Li","year":"2024"},{"key":"ref45","volume-title":"Innovations in Electrical and Electronic Engineering. Lecture Notes in Electrical Engineering","author":"Joseph","year":"2021"},{"key":"ref46","doi-asserted-by":"publisher","first-page":"1162","DOI":"10.54097\/47fwzy91","article-title":"A CNN-based implementation of fruit recognition","volume":"85","author":"Huang","year":"2024","journal-title":"Highl Sci Eng Technol"},{"key":"ref47","doi-asserted-by":"publisher","first-page":"20251223","DOI":"10.31584\/jhsmr.20251223","article-title":"Mobile app engagement and effective weight management","volume":"44","author":"Surangsrirat","year":"2024","journal-title":"J Health Sci Med Res"}],"container-title":["Frontiers in Public Health"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fpubh.2026.1875869\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,8,25]],"date-time":"2026-08-25T05:34:50Z","timestamp":1787636090000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fpubh.2026.1875869\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,8,25]]},"references-count":47,"alternative-id":["10.3389\/fpubh.2026.1875869"],"URL":"https:\/\/doi.org\/10.3389\/fpubh.2026.1875869","relation":{},"ISSN":["2296-2565"],"issn-type":[{"value":"2296-2565","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,8,25]]},"article-number":"1875869"}}