{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,9,7]],"date-time":"2026-09-07T18:30:26Z","timestamp":1788805826312,"version":"build-2803163510"},"reference-count":34,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2026,4,21]],"date-time":"2026-04-21T00:00:00Z","timestamp":1776729600000},"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. Plant Sci."],"abstract":"<jats:sec>\n                    <jats:title>Introduction<\/jats:title>\n                    <jats:p>Cashew apple is a nutrient-rich fruit containing abundant minerals, vitamins, and energy. However, its fleshy texture and delicate skin significantly limit its storage life and market value. Accurate maturity grading is therefore essential for improving post-harvest management and transportation efficiency.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>This study proposes a lightweight vision transformer (ViT) student model trained using multi-granular knowledge distillation (KD) from a stronger data-efficient image transformer (DeiT)-Base teacher. The distillation framework integrates response-based soft-label supervision, attention transfer, and token-level feature regression to enhance representation learning under limited data conditions. Auxiliary lightweight architectures, including MobileNet, ConvNeXt, and EdgeNeXt, were trained independently to provide complementary predictions, and a weighted fusion strategy was employed for ensemble evaluation.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>The proposed ensemble ViT-KD with EdgeNeXt achieved 90% accuracy under the evaluated test split. To ensure statistical reliability and address potential partition bias, a stratified fivefold cross-validation was conducted on the dataset, yielding a mean accuracy of 86.89% \u00b1 2.89% with consistent F1 scores and recall. The relatively low variance across the folds indicates stable internal generalization. Comparative experiments with conventional convolutional neural network (CNN) baselines and lightweight CNN baselines such as MobileViT-S and ShuffleNetV2 were performed, with the proposed ensemble framework achieving improved accuracy while maintaining computational efficiency. Computational analysis indicates that the stand-alone distilled ViT maintains a real-time inference capability of 8.79 ms per image, which supports suitability for edge-oriented agricultural applications.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Discussion<\/jats:title>\n                    <jats:p>These results highlight the effectiveness of knowledge-distilled lightweight transformers for data-efficient maturity grading of cashew apples.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.3389\/fpls.2026.1787609","type":"journal-article","created":{"date-parts":[[2026,4,21]],"date-time":"2026-04-21T05:35:59Z","timestamp":1776749759000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":1,"title":["Distilled vision transformers with CNN fusion for robust cashew apple maturity prediction"],"prefix":"10.3389","volume":"17","author":[{"given":"Sumalatha","family":"Lingamgunta","sequence":"first","affiliation":[{"name":"Department of Computer Science and Engineering, University College of Engineering Kakinada, Jawaharlal Nehru Technological University Kakinada","place":["Kakinada, Andhra Pradesh, India"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Jeevaratnam","family":"Mudidana","sequence":"additional","affiliation":[{"name":"Department of Computer Science and Engineering, University College of Engineering Kakinada, Jawaharlal Nehru Technological University Kakinada","place":["Kakinada, Andhra Pradesh, India"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Durai Raj","family":"Vincent","sequence":"additional","affiliation":[{"name":"School of Computer Science Engineering and Information Systems, Vellore Institute of Technology","place":["Vellore, India"]}],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Felicita","family":"S. A. M.","sequence":"additional","affiliation":[{"name":"School of Computer Science Engineering and Information Systems, Vellore Institute of Technology","place":["Vellore, India"]}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"1965","published-online":{"date-parts":[[2026,4,21]]},"reference":[{"key":"B1","doi-asserted-by":"publisher","DOI":"10.3390\/electronics11244100","article-title":"Prediction of fruit maturity, quality, and its life using deep learning algorithms","volume":"11","author":"Aherwadi","year":"2022","journal-title":"Electronics"},{"key":"B2","doi-asserted-by":"publisher","DOI":"10.7717\/peerj-cs.474","article-title":"Knowledge distillation in deep learning and its applications","volume":"7","author":"Alkhulaifi","year":"2021","journal-title":"PeerJ Comput. Sci."},{"key":"B3","doi-asserted-by":"publisher","DOI":"10.1016\/j.foodres.2024.114836","article-title":"Interpretation and explanation of computer vision classification of carambola (Averrhoa carambola L.) according to maturity stage","volume":"192","author":"Alves de Moraes","year":"2024","journal-title":"Food Res. Int."},{"key":"B4","article-title":"Electronic nose and deep learning approach in identifying ripe Lycopersicum esculentum L. tomato fruit","author":"Anticuando","year":"2022"},{"key":"B5","first-page":"1","article-title":"Automatic cashew detection system using deep learning","author":"Charukeshi","year":"2024"},{"key":"B6","doi-asserted-by":"publisher","DOI":"10.1016\/j.atech.2025.101333","article-title":"Assessing deep learning model robustness for banana ripeness classification under varying illumination conditions","volume":"12","author":"Chuquimarca","year":"2025","journal-title":"Smart Agric. Technol."},{"key":"B7","doi-asserted-by":"publisher","first-page":"250","DOI":"10.1186\/s40537-025-01296-2","article-title":"Artificial intelligence advances for cashew fruit maturity and quality detection: a systematic review on models, sensors, and farming applications","volume":"12","author":"Colaco","year":"2025","journal-title":"J. Big Data"},{"key":"B8","doi-asserted-by":"publisher","first-page":"87","DOI":"10.1016\/j.jfoodeng.2016.09.011","article-title":"Post-harvest processing technology for cashew apple\u2013a review","volume":"194","author":"Das","year":"2017","journal-title":"J. Food Eng."},{"key":"B9","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2010.11929","article-title":"An image is worth 16x16 words: Transformers for image recognition at scale","author":"Dosovitskiy","year":"2020","journal-title":"arXiv preprint arXiv:201011929"},{"key":"B10","doi-asserted-by":"publisher","DOI":"10.1016\/j.jfutfo.2025.02.001","article-title":"Classification of strawberry maturity using Swin Transformer V2: Integrating near-infrared spectral, chemical, and optical properties","author":"Feng","year":"2025","journal-title":"J. Future Foods"},{"key":"B11","doi-asserted-by":"publisher","first-page":"2715","DOI":"10.1007\/s11694-023-02349-x","article-title":"Tomato ripeness and shelf-life prediction system using machine learning","volume":"18","author":"Goyal","year":"2024","journal-title":"J. Food Meas. Charact."},{"key":"B12","doi-asserted-by":"publisher","first-page":"170","DOI":"10.1016\/j.biosystemseng.2019.04.018","article-title":"An adaptable approach to automated visual detection of plant organs with applications in grapevine breeding","volume":"183","author":"Grimm","year":"2019","journal-title":"Biosyst. Eng."},{"key":"B13","doi-asserted-by":"publisher","first-page":"398","DOI":"10.20944\/preprints202510.1697.v1","article-title":"Assessment of physicochemical properties of cashew apple through computer vision","volume":"7","author":"Gupta","year":"2025","journal-title":"AgriEngineering"},{"key":"B14","doi-asserted-by":"publisher","DOI":"10.48550\/arXiv.2502.20223","article-title":"Deep convolutional neural networks for palm fruit maturity classification","author":"Han","year":"2025","journal-title":"arXiv:2502.20223v1 [cs.CV]"},{"key":"B15","first-page":"766","article-title":"Improvement of E-Nose sensor signal using MVA, FFT, DWT methods on pineapple fruit maturity","author":"Hasan","year":"2022"},{"key":"B16","doi-asserted-by":"publisher","first-page":"95496","DOI":"10.1109\/access.2025.3571735","article-title":"Transformers for vision: A survey on innovative methods for computer vision","volume":"13","author":"Hassija","year":"2025","journal-title":"IEEE Access"},{"key":"B17","doi-asserted-by":"publisher","first-page":"70","DOI":"10.1016\/j.compag.2018.02.016","article-title":"Deep learning in agriculture: A survey","volume":"147","author":"Kamilaris","year":"2018","journal-title":"Comput. Electron. Agric."},{"key":"B18","first-page":"3","article-title":"Edgenext: efficiently amalgamated cnn-transformer architecture for mobile vision applications","author":"Maaz","year":"2022"},{"key":"B19","doi-asserted-by":"crossref","DOI":"10.1109\/IICAIET55139.2022.9936870","article-title":"Determination of tomato fruit stages using principal component analysis and fuzzy logic algorithm","author":"MacAtangay","year":"2022"},{"key":"B20","doi-asserted-by":"publisher","DOI":"10.1016\/j.atech.2022.100099","article-title":"Detection of fruit maturity stage and yield estimation in wild blueberry using deep learning convolutional neural networks","volume":"3","author":"MacEachern","year":"2023","journal-title":"Smart Agric. Technol."},{"key":"B21","doi-asserted-by":"publisher","DOI":"10.3390\/pr13071982","article-title":"Artificial vision-based dual CNN classification of banana ripeness and quality attributes using RGB images","volume":"13","author":"Mart\u00ednez-Mora","year":"2025","journal-title":"Processes"},{"key":"B22","doi-asserted-by":"publisher","DOI":"10.3389\/fpls.2019.00545","article-title":"Fruit ripening: From present knowledge to future development","volume":"10","author":"Palma","year":"2019","journal-title":"Front. Plant Sci."},{"key":"B23","first-page":"1742","article-title":"Weakly-and semi-supervised learning of a deep convolutional network for semantic image segmentation","author":"Papandreou","year":"2015"},{"key":"B24","article-title":"Goa cashew apple maturity grading","volume-title":"IEEE dataport","author":"Sawant","year":"2025"},{"key":"B25","doi-asserted-by":"publisher","DOI":"10.1016\/j.atech.2022.100130","article-title":"Fruit maturity grading framework for small dataset using single image multi-object sampling and Mask R-CNN","volume":"3","author":"Siricharoen","year":"2023","journal-title":"Smart Agric. Technol."},{"key":"B26","doi-asserted-by":"publisher","DOI":"10.3390\/app132212504","article-title":"Deep learning-based method for classification and ripeness assessment of fruits and vegetables","volume":"13","author":"Tapia-Mendez","year":"2023","journal-title":"Appl. Sci."},{"key":"B27","first-page":"10347","article-title":"Training data-efficient image transformers & distillation through attention","author":"Touvron","year":"2021"},{"key":"B28","doi-asserted-by":"publisher","DOI":"10.1016\/j.atech.2023.100394","article-title":"Maturity stages detection prototype device for classifying custard apple (Annona squamosa L) fruit using image processing approach","volume":"7","author":"Wakchaure","year":"2024","journal-title":"Smart Agric. Technol."},{"key":"B29","doi-asserted-by":"publisher","DOI":"10.3389\/fpls.2022.868745","article-title":"Application of convolutional neural network-based detection methods in fresh fruit production: a comprehensive review","volume":"13","author":"Wang","year":"2022","journal-title":"Front. Plant Sci."},{"key":"B30","doi-asserted-by":"publisher","DOI":"10.3390\/app142311111","article-title":"A method for detecting tomato maturity based on deep learning","volume":"14","author":"Wang","year":"2024","journal-title":"Appl. Sci."},{"key":"B31","doi-asserted-by":"publisher","DOI":"10.1371\/journal.pone.0326103","article-title":"A deep learning based approach for classifying the maturity of cashew apples","volume":"20","author":"Winklmair","year":"2025","journal-title":"PloS One"},{"key":"B32","doi-asserted-by":"publisher","first-page":"7811","DOI":"10.1007\/s11042-023-15938-1","article-title":"Apple ripeness identification from digital images using transformers","volume":"83","author":"Xiao","year":"2024","journal-title":"Multimed Tools Appl."},{"key":"B33","first-page":"2571","article-title":"Detection of single grapevine berries in images using fully convolutional neural networks","author":"Zabawa","year":"2019"},{"key":"B34","doi-asserted-by":"publisher","first-page":"107499","DOI":"10.1016\/j.neunet.2025.107499","article-title":"MSCViT: A small-size ViT architecture with multi-scale self-attention mechanism for tiny datasets","volume":"188","author":"Zhang","year":"2025","journal-title":"Neural Networks"}],"container-title":["Frontiers in Plant Science"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fpls.2026.1787609\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,4,21]],"date-time":"2026-04-21T05:36:05Z","timestamp":1776749765000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fpls.2026.1787609\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,4,21]]},"references-count":34,"alternative-id":["10.3389\/fpls.2026.1787609"],"URL":"https:\/\/doi.org\/10.3389\/fpls.2026.1787609","relation":{},"ISSN":["1664-462X"],"issn-type":[{"value":"1664-462X","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,4,21]]},"article-number":"1787609"}}