{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,25]],"date-time":"2026-09-25T18:00:29Z","timestamp":1790359229164,"version":"4.1.0"},"reference-count":33,"publisher":"Public Library of Science (PLoS)","issue":"9","license":[{"start":{"date-parts":[[2026,9,25]],"date-time":"2026-09-25T00:00:00Z","timestamp":1790294400000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["www.plosone.org"],"crossmark-restriction":false},"short-container-title":["PLoS One"],"abstract":"<jats:p>Accurate recognition of food images is essential for image-based dietary assessment, nutritional tracking, and digital health applications. However, closed-set Chinese dish recognition remains challenging because many categories in the CNFOOD-241 dataset exhibit strong visual similarity, substantial intra-class variation, and subtle inter-class differences. In this study, we present an integrated discriminative training strategy for fine-grained Chinese dish recognition using RegNetY-32GF as the student backbone. Rather than proposing a new backbone architecture, the study focuses on improving category-level discrimination under a fixed convolutional architecture by combining teacher-guided knowledge distillation, margin-based classification, triplet-based embedding regularization, and mixed-sample augmentation. Across five random seeds, the proposed full model achieved 84.37\u2009+\u2009\/- 0.29% Top-1 accuracy, 97.66\u2009+\u2009\/- 0.11% Top-5 accuracy, 83.41\u2009+\u2009\/- 0.57% macro-F1, and 84.27\u2009+\u2009\/- 0.30% weighted-F1. Paired bootstrap analysis showed a positive Top-1 improvement of 0.651 percentage points over the baseline, with a 95% confidence interval of [0.491, 0.817], and the mean class-wise Delta F1 across five seeds was significantly greater than zero according to the Wilcoxon signed-rank test (p\u2009=\u20090.017803). These findings suggest that the proposed framework provides a modest improvement in category-level discrimination on CNFOOD-241, with statistically supported evidence at the dataset level, although the magnitude of improvement varies across individual categories. The training code, evaluation scripts, derived result files, and trained model weights are publicly available at the project repository and GitHub Release.<\/jats:p>","DOI":"10.1371\/journal.pone.0358248","type":"journal-article","created":{"date-parts":[[2026,9,25]],"date-time":"2026-09-25T17:28:03Z","timestamp":1790357283000},"page":"e0358248","update-policy":"https:\/\/doi.org\/10.1371\/journal.pone.corrections_policy","source":"Crossref","is-referenced-by-count":0,"title":["Fine-grained Chinese dish recognition on CNFOOD-241: A discriminative learning approach for improving category-level discrimination"],"prefix":"10.1371","volume":"21","author":[{"given":"Xin","family":"Xiong","sequence":"first","affiliation":[{"name":"Kunming University of Science and Technology","place":["Kunming, China"],"department":["Faculty of Information Engineering and Automation"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0009-0008-7670-0915","authenticated-orcid":true,"given":"Jinqiang","family":"Liao","sequence":"additional","affiliation":[{"name":"Kunming University of Science and Technology","place":["Kunming, China"],"department":["Faculty of Information Engineering and Automation"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Xiao","family":"Mo","sequence":"additional","affiliation":[{"name":"Kunming University of Science and Technology","place":["Kunming, China"],"department":["Faculty of Information Engineering and Automation"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"JianFeng","family":"He","sequence":"additional","affiliation":[{"name":"Kunming University of Science and Technology","place":["Kunming, China"],"department":["Faculty of Information Engineering and Automation"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"340","published-online":{"date-parts":[[2026,9,25]]},"reference":[{"issue":"5","key":"pone.0358248.ref001","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1145\/3329168","article-title":"A Survey on Food Computing","volume":"52","author":"W Min","year":"2019","journal-title":"ACM Comput Surv"},{"key":"pone.0358248.ref002","first-page":"446","author":"L Bossard","year":"2014","journal-title":"Food-101\u2014Mining discriminative components with random forests. In: Computer Vision - ECCV 2014. Cham: Springer"},{"issue":"3","key":"pone.0358248.ref003","doi-asserted-by":"crossref","first-page":"588","DOI":"10.1109\/JBHI.2016.2636441","article-title":"Food recognition: A new dataset, experiments, and results","volume":"21","author":"G Ciocca","year":"2017","journal-title":"IEEE J Biomed Health Inform"},{"issue":"7","key":"pone.0358248.ref004","doi-asserted-by":"crossref","first-page":"657","DOI":"10.3390\/nu9070657","article-title":"NutriNet: A deep learning food and drink image recognition system for dietary assessment","volume":"9","author":"S Mezgec","year":"2017","journal-title":"Nutrients"},{"issue":"12","key":"pone.0358248.ref005","doi-asserted-by":"crossref","first-page":"8927","DOI":"10.1109\/TPAMI.2021.3126648","article-title":"Fine-grained image analysis with deep learning: A survey","volume":"44","author":"X-S Wei","year":"2022","journal-title":"IEEE Trans Pattern Anal Mach Intell"},{"key":"pone.0358248.ref006","first-page":"1449","author":"T-Y Lin","year":"2015"},{"key":"pone.0358248.ref007","first-page":"4476","author":"J Fu","year":"2017"},{"key":"pone.0358248.ref008","first-page":"1","article-title":"Automatic Chinese Food recognition based on a stacking fusion model","volume":"2023","author":"B Fan","year":"2023","journal-title":"Annu Int Conf IEEE Eng Med Biol Soc"},{"key":"pone.0358248.ref009","doi-asserted-by":"crossref","first-page":"101733","DOI":"10.1016\/j.jafr.2025.101733","article-title":"Large-scale image classification and nutrient estimation for Chinese dishes","volume":"19","author":"Y Feng","year":"2025","journal-title":"J Agriculture Food Res"},{"issue":"5","key":"pone.0358248.ref010","article-title":"Improving fine-grained food classification using deep residual learning and selective state space models","volume":"20","author":"C-S Chen","year":"2025","journal-title":"PLoS One"},{"issue":"8","key":"pone.0358248.ref011","doi-asserted-by":"crossref","first-page":"9932","DOI":"10.1109\/TPAMI.2023.3237871","article-title":"Large Scale Visual Food Recognition","volume":"45","author":"W Min","year":"2023","journal-title":"IEEE Trans Pattern Anal Mach Intell"},{"key":"pone.0358248.ref012","first-page":"3308","author":"M Gwilliam","year":"2021"},{"issue":"4","key":"pone.0358248.ref013","doi-asserted-by":"crossref","first-page":"427","DOI":"10.1016\/j.ipm.2009.03.002","article-title":"A systematic analysis of performance measures for classification tasks","volume":"45","author":"M Sokolova","year":"2009","journal-title":"Inf Process Manag"},{"key":"pone.0358248.ref014","first-page":"9260","author":"Y Cui","year":"2019"},{"key":"pone.0358248.ref015","first-page":"10425","author":"I Radosavovic","year":"2020"},{"key":"pone.0358248.ref016","doi-asserted-by":"crossref","first-page":"1789","DOI":"10.1007\/s11263-021-01453-z","article-title":"Knowledge distillation: A survey","volume":"129","author":"J Gou","year":"2021","journal-title":"Int J Comput Vis"},{"key":"pone.0358248.ref017","first-page":"4685","author":"J Deng","year":"2019"},{"key":"pone.0358248.ref018","first-page":"770","author":"K He","year":"2016"},{"key":"pone.0358248.ref019","first-page":"2261","author":"G Huang","year":"2017"},{"key":"pone.0358248.ref020","first-page":"5987","author":"S Xie","year":"2017"},{"key":"pone.0358248.ref021","unstructured":"Tan M, Le QV. EfficientNet: Rethinking model scaling for convolutional neural networks. In: Proceedings of the 36th International Conference on Machine Learning, 2019. 6105\u201314."},{"key":"pone.0358248.ref022","author":"W Park","year":"2019"},{"key":"pone.0358248.ref023","first-page":"9155","author":"S Ahn","year":"2019"},{"key":"pone.0358248.ref024","unstructured":"Liu W, Wen Y, Yu Z, Yang M. Large-margin softmax loss for convolutional neural networks. In: Proceedings of the 33rd International Conference on Machine Learning, 2016. 507\u201316."},{"key":"pone.0358248.ref025","first-page":"5265","author":"H Wang","year":"2018"},{"key":"pone.0358248.ref026","author":"R Hadsell","year":"2006"},{"key":"pone.0358248.ref027","first-page":"815","author":"F Schroff","year":"2015"},{"key":"pone.0358248.ref028","unstructured":"Sohn K. Improved deep metric learning with multi-class N-pair loss objective. In: Advances in Neural Information Processing Systems. 2016."},{"key":"pone.0358248.ref029","unstructured":"Zhang H, Cisse M, Dauphin YN, Lopez-Paz D. Mixup: beyond empirical risk minimization. In: International Conference on Learning Representations (ICLR). 2018."},{"key":"pone.0358248.ref030","first-page":"6022","author":"S Yun","year":"2019"},{"key":"pone.0358248.ref031","doi-asserted-by":"crossref","first-page":"130994","DOI":"10.1016\/j.foodchem.2021.130994","article-title":"Application of deep learning for image-based Chinese market food nutrients estimation","volume":"373","author":"P Ma","year":"2022","journal-title":"Food Chem"},{"issue":"6","key":"pone.0358248.ref032","doi-asserted-by":"crossref","first-page":"2480","DOI":"10.1109\/TCSVT.2020.3020079","article-title":"Food and Ingredient Joint Learning for Fine-Grained Recognition","volume":"31","author":"C Liu","year":"2021","journal-title":"IEEE Trans Circuits Syst Video Technol"},{"key":"pone.0358248.ref033","first-page":"32","author":"J Chen","year":"2016"}],"container-title":["PLOS One"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pone.0358248","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,9,25]],"date-time":"2026-09-25T17:28:10Z","timestamp":1790357290000},"score":1,"resource":{"primary":{"URL":"https:\/\/dx.plos.org\/10.1371\/journal.pone.0358248"}},"subtitle":[],"editor":[{"given":"Toqir","family":"Rana","sequence":"first","affiliation":[{"name":"The University of Lahore","place":["PAKISTAN"]}],"role":[{"vocabulary":"crossref","role":"editor"}]}],"short-title":[],"issued":{"date-parts":[[2026,9,25]]},"references-count":33,"journal-issue":{"issue":"9","published-online":{"date-parts":[[2026,9,25]]}},"URL":"https:\/\/doi.org\/10.1371\/journal.pone.0358248","relation":{},"ISSN":["1932-6203"],"issn-type":[{"value":"1932-6203","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,9,25]]}}}