tatpipТехника и технология пищевых производствFood Processing: Techniques and Technology2074-94142313-1748Кемеровский государственный университет10.21603/2074-9414-2026-3-2657ALMTCWtatpip-251Research ArticleСтатьиОпределение товарного сорта ягод вишни с использованием сверточных нейронных сетейConvolutional Neural Networks in Commercial Grading of Cherryhttps://orcid.org/0000-0003-0728-7211ПоповА. М.PopovA. M.

Попов Анатолий Михайлович

Кемерово

Anatoly M. Popov

Kemerovo

https://orcid.org/0000-0003-4512-1933ШафрайА. В.ShafraiA. V.

Шафрай Антон Валерьевич

Кемерово

Anton V. Shafrai

Kemerovo

shafraia@mail.ruhttps://orcid.org/0009-0005-5852-9809КосиновВ. С.KosinovV. S.

Косинов Виталий Сергеевич

Кемерово

Vitaly S. Kosinov

Kemerovo

https://orcid.org/0000-0003-4499-8426ПодберезенГ. И.PodberezenG. I.

Подберезен Григорий Игоревич

Кемерово

Grigory I. Podberezen

Kemerovo

https://orcid.org/0000-0001-7995-3813СухоруковД. В.SukhorukovD. V.

Сухоруков Дмитрий Викторович

Кемерово

Dmitry V. Sukhorukov

Kemerovo

https://orcid.org/0000-0002-1517-1596КомаровС. С.KomarovS. S.

Комаров Сергей Сергеевич

Кемерово

Sergey S. Komarov

Kemerovo

Кемеровский государственный университетРоссияKemerovo State UniversityRussian Federation202606102026563516525Copyright © Попов А.М., Шафрай А.В., Косинов В.С., Подберезен Г.И., Сухоруков Д.В., Комаров С.С., 20262026Попов А.М., Шафрай А.В., Косинов В.С., Подберезен Г.И., Сухоруков Д.В., Комаров С.С.Popov A.M., Shafrai A.V., Kosinov V.S., Podberezen G.I., Sukhorukov D.V., Komarov S.S.Данная работа распространяется под лицензией Creative Commons Attribution 4.0.This work is licensed under a Creative Commons Attribution 4.0 License.https://www.fptt.ru/jour/article/view/251

Ручной контроль качества плодово-ягодного сырья на этапе приемки остается одной из ключевых задач пищевой промышленности. Существующие методы визуальной оценки субъективны, трудоемки и зависят от квалификации оператора, что ограничивает их эффективность в условиях современного производства. Цель исследования – применить модели сверточных нейронных сетей семейства YOLO для классификации ягод вишни по товарным сортам на этапе приемки сырья. Объект исследования – ягоды вишни культурных сортов (Prunus subg. Cerasus), собранные в Кемеровской области в 2024 г. (Россия). Исследование выполнено на базе ООО «НПО Здоровое питание» (Кемерово, Россия). Собраны и размечены изображения четырех категорий качества. Проведено сравнение пяти моделей YOLO (YOLOv8m, YOLOv9m, YOLO11m, YOLO12m, YOLO26m) и трех вариантов архитектуры YOLO12 по размеру. Для выбранной модели YOLO12m оптимизированы гиперпараметры: сравнивались восемь оптимизаторов, четыре значения размера батча и пять значений коэффициента регуляризации Dropout.Установлено, что наилучшую точность показала модель среднего размера YOLO12. Среди оптимизаторов наивысшие метрики зафиксированы для SGD. Определен оптимальный размер обучающего пакета; изменение параметра Dropout не оказало заметного влияния на результаты. Увеличение числа эпох финального обучения повысило качество распознавания по всем четырем категориям. Работоспособность модели подтверждена на производственной выборке. На основе модели разработан прототип программы с графическим интерфейсом для загрузки изображений, визуализации детекции и формирования отчета.Результаты подтверждают пригодность моделей YOLO для автоматизации входного контроля качества ягод вишни, обес- печивая объективный и воспроизводимый анализ взамен субъективной ручной оценки, и создают основу для внедрения промышленных систем контроля качества.

Manual quality control of fruit and berries remains a major challenge in the food industry. Existing visual assessment methods are subjective, labor-intensive, and dependent on operator skills, which limits their effectiveness in modern production environments. However, convolutional neural network models of the YOLO type can be used to classify cherries by commercial grade at the raw material acceptance stage.The study focused on cultivated cherries (Prunus subg. Cerasus) harvested in the Kemerovo Region (Russia) in 2024. The research was conducted at the Zdorovye Pitanie Research and Production Association (Kemerovo, Russia). The comparative analysis evaluated images across four quality categories using five YOLO models (YOLOv8m, YOLOv9m, YOLO11m, YOLO12m, YOLO26m) and three YOLO12 architectural variants of different sizes. Hyperparameter optimization was performed for the selected YOLO12m model, assessing eight optimizers, four batch sizes, and five Dropout regularization coefficient values.The medium-sized YOLO12 model demonstrated the highest accuracy. Among the optimizers, SGD yielded the best performance metrics. The optimal training batch size was determined, while adjusting the Dropout parameter had no significant effect on the results. Increasing the number of final training epochs improved recognition performance across all four categories. The model’s performance was validated on an industrial sample. Based on this model, a prototype software application with a graphical user interface (GUI) was developed for image loading, detection visualization, and report generation.The results confirm the suitability of YOLO models for automating incoming cherry quality control, providing objective and reproducible analysis instead of subjective manual assessment, thereby paving the way for advanced industrial quality control systems.

Искусственный интеллектнейронные сетиYOLOпищевая промышленностьпереработка сырьяягоды вишниArtificial intelligenceneural networksYOLOfood industryraw material processingcherryИсследование выполнено за счет гранта Российского научного фонда № 25-26-00136, https://rscf.ru/project/25-26-00136/The study was supported by the Russian Science Foundation, Grant No. 25-26-00136, https://rscf.ru/en/project/25-26-00136/ReferencesШафрай А. В., Попов А. М., Косинов В. С., Резниченко И. Ю., Бондарчук О. Н. Применение сверточных ней- ронных сетей для контроля показателей технологического потока производства гранулированных киселей. Техника и технология пищевых производств. 2025. Т. 55. № 4. С. 845–855. https://doi.org/10.21603/2074-9414-2025-4-2613Shafrai AV, Popov AM, Kosinov VS, Reznichenko IYu, Bondarchuk ON. Convolutional neural networks in granulated kissel production. Food Processing: Techniques and Technology. 2025;55(4):845–855. (In Russ.) https://doi.org/10.21603/2074-9414-2025-4-2613Daiaeddine MJ, Badrouss S, El Harti A, Bachaoui EM, Biniz M, et al. UAV imagery, advanced deep learning, and YOLOv7 object detection model in enhancing citrus yield estimation. Foods and Raw Materials. 2025;13(2):242–253. https://doi.org/10.21603/2308-4057-2025-2-650Daiaeddine MJ, Badrouss S, El Harti A, Bachaoui EM, Biniz M, et al. UAV imagery, advanced deep learning, and YOLOv7 object detection model in enhancing citrus yield estimation. Foods and Raw Materials. 2025;13(2):242–253. https://doi.org/10.21603/2308-4057-2025-2-650Yavuzer E, Köse M, Uslu H. Determining the quality level of ready to-eat stuffed mussels with Arduino-based electronic nose. Journal of Food Measurement and Characterization. 2024;18:5629–5637. https://doi.org/10.1007/s11694-024-02593-9Yavuzer E, Köse M, Uslu H. Determining the quality level of ready to-eat stuffed mussels with Arduino-based electronic nose. Journal of Food Measurement and Characterization. 2024;18:5629–5637. https://doi.org/10.1007/s11694-024-02593-9Wijaya DR, Syarwan NF, Nugraha MA, Ananda D, Fahrudin T, et al. Seafood quality detection using electronic nose and machine learning algorithms with hyperparameter optimization. IEEE Access. 2023;11:62484–62495. https://doi.org/ 10.1109/ACCESS.2023.3286980Wijaya DR, Syarwan NF, Nugraha MA, Ananda D, Fahrudin T, et al. Seafood quality detection using electronic nose and machine learning algorithms with hyperparameter optimization. IEEE Access. 2023;11:62484–62495. https://doi.org/ 10.1109/ACCESS.2023.3286980Zlobin I, Sinelnikov A, Toroptsev N, Titov K, Antoshkina E, et al. Metal–organic frameworks at the tip of the e-tongue: Machine learning-driven disposable electrochemical sensors. Journal of Materials Chemistry C. 2025;13(23):11776–11788. https://doi.org/10.1039/D5TC00965KZlobin I, Sinelnikov A, Toroptsev N, Titov K, Antoshkina E, et al. Metal–organic frameworks at the tip of the e-tongue: Machine learning-driven disposable electrochemical sensors. Journal of Materials Chemistry C. 2025;13(23):11776–11788. https://doi.org/10.1039/D5TC00965KAnisimov DS, Abramov AA, Gaidarzhi VP, Kaplun DS, Agina EV, et al. Food freshness measurements and product distinguishing by a portable electronic nose based on organic field-effect transistors. ACS Omega. 2023;8(5):4649–4654. https:// doi.org/10.1021/acsomega.2c06386Anisimov DS, Abramov AA, Gaidarzhi VP, Kaplun DS, Agina EV, et al. Food freshness measurements and product distinguishing by a portable electronic nose based on organic field-effect transistors. ACS Omega. 2023;8(5):4649–4654. https:// doi.org/10.1021/acsomega.2c06386Shuba A, Umarkhanov R, Bogdanova E, Anokhina E, Burakova I. Possibilities of an electronic nose on piezoelectric sensors with polycomposite coatings to investigate the microbiological indicators of milk. Sensors. 2024;24(11):3634. https://doi.org/10.3390/s24113634Shuba A, Umarkhanov R, Bogdanova E, Anokhina E, Burakova I. Possibilities of an electronic nose on piezoelectric sensors with polycomposite coatings to investigate the microbiological indicators of milk. Sensors. 2024;24(11):3634. https://doi.org/10.3390/s24113634Fedorov FS, Yaqin A, Krasnikov DV, Kondrashov VA, Ovchinnikov G, et al. Detecting cooking state of grilled chicken by electronic nose and computer vision techniques. Food Chemistry. 2021;345:128747. https://doi.org/10.1016/j.food-chem.2020.128747Fedorov FS, Yaqin A, Krasnikov DV, Kondrashov VA, Ovchinnikov G, et al. Detecting cooking state of grilled chicken by electronic nose and computer vision techniques. Food Chemistry. 2021;345:128747. https://doi.org/10.1016/j.food-chem.2020.128747Wu X, Xu B, Ma R, Niu Y, Gao S, et al. Identification and quantification of adulterated honey by Raman spectroscopy combined with convolutional neural network and chemometrics. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy. 2022;274(1):121133. https://doi.org/10.1016/j.saa.2022.121133Wu X, Xu B, Ma R, Niu Y, Gao S, et al. Identification and quantification of adulterated honey by Raman spectroscopy combined with convolutional neural network and chemometrics. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy. 2022;274(1):121133. https://doi.org/10.1016/j.saa.2022.121133Lanjewar MG, Asolkar S, Parab JS, Morajkar PP. Detecting starch-adulterated turmeric using Vis-NIR spectroscopy and multispectral imaging with machine learning. Journal of Food Composition and Analysis. 2024;136:106700. https://doi.org/10.1016/j.jfca.2024.106700Lanjewar MG, Asolkar S, Parab JS, Morajkar PP. Detecting starch-adulterated turmeric using Vis-NIR spectroscopy and multispectral imaging with machine learning. Journal of Food Composition and Analysis. 2024;136:106700. https://doi.org/10.1016/j.jfca.2024.106700Saha D, Senthilkumar T, Singh CHB, Manickavasagan A. Quantitative detection of metanil yellow adulteration in chickpea flour using line-scan near-infrared hyperspectral imaging with partial least square regression and one-dimensional convolutional neural network. Journal of Food Composition and Analysis. 2023;120:105290. https://doi.org/10.1016/j.jfca.2023.105290Saha D, Senthilkumar T, Singh CHB, Manickavasagan A. Quantitative detection of metanil yellow adulteration in chickpea flour using line-scan near-infrared hyperspectral imaging with partial least square regression and one-dimensional convolutional neural network. Journal of Food Composition and Analysis. 2023;120:105290. https://doi.org/10.1016/j.jfca.2023.105290Belugina R, Senchikhina A, Volkov S, Fedorov A, Legin A, et al. Quantification of phosphatides in sunflower oils using a potentiometric e-tongue. Analytical Methods. 2022;14(32):3064–3070. https://doi.org/10.1039/D2AY00736CBelugina R, Senchikhina A, Volkov S, Fedorov A, Legin A, et al. Quantification of phosphatides in sunflower oils using a potentiometric e-tongue. Analytical Methods. 2022;14(32):3064–3070. https://doi.org/10.1039/D2AY00736CSánchez CN, Orvañanos-Guerrero MT, Domínguez-Soberanes JU, Álvarez-Cisneros YEM. Analysis of beef quality according to color changes using computer vision and white-box machine learning techniques. Heliyon. 2023;9(7):e17976. https://doi.org/10.1016/j.heliyon.2023.e17976Sánchez CN, Orvañanos-Guerrero MT, Domínguez-Soberanes JU, Álvarez-Cisneros YEM. Analysis of beef quality according to color changes using computer vision and white-box machine learning techniques. Heliyon. 2023;9(7):e17976. https://doi.org/10.1016/j.heliyon.2023.e17976Huang W, Wang X, Zhang JU, Xia J, Zhang X. Improvement of blueberry freshness prediction based on machine learning and multi-source sensing in the cold chain logistics. Food Control. 2023;145:109496. https://doi.org/10.1016/j.food-cont.2022.109496Huang W, Wang X, Zhang JU, Xia J, Zhang X. Improvement of blueberry freshness prediction based on machine learning and multi-source sensing in the cold chain logistics. Food Control. 2023;145:109496. https://doi.org/10.1016/j.food-cont.2022.109496Anwar H, Anwar T, Murtaza SH. Review on food quality assessment using machine learning and electronic nose system. Biosensors and Bioelectronics. 2023;14:100365. https://doi.org/10.1016/j.biosx.2023.100365Anwar H, Anwar T, Murtaza SH. Review on food quality assessment using machine learning and electronic nose system. Biosensors and Bioelectronics. 2023;14:100365. https://doi.org/10.1016/j.biosx.2023.100365Ксенофонтов В. В. Нейронные сети. Проблемы науки. 2020. № 11. С. 28–29. [Ksenofontov VV. Neural networks. Problems of Science. 2020;(11):28–29. (In Russ.)] https://elibrary.ru/DTVJNSКсенофонтов В. В. Нейронные сети. Проблемы науки. 2020. № 11. С. 28–29. [Ksenofontov VV. Neural networks. Problems of Science. 2020;(11):28–29. (In Russ.)] https://elibrary.ru/DTVJNSSapkota R, Flores-Calero M, Qureshi R, Badgujar CH, Nepal U. YOLO advances to its genesis: A decadal and comprehensive review of the You Only Look Once (YOLO) series. Artificial Intelligence Review. 2025;58(9):274. https://doi.org/10.1007/s10462-025-11253-3Sapkota R, Flores-Calero M, Qureshi R, Badgujar CH, Nepal U. YOLO advances to its genesis: A decadal and comprehensive review of the You Only Look Once (YOLO) series. Artificial Intelligence Review. 2025;58(9):274. https://doi.org/10.1007/s10462-025-11253-3Mirzaei B, Nezamabadi-Pour H, Raoof A, Derakhshani R. Small object detection and tracking: A comprehensive review. Sensors. 2023;23(15):6887. https://doi.org/10.3390/s23156887Mirzaei B, Nezamabadi-Pour H, Raoof A, Derakhshani R. Small object detection and tracking: A comprehensive review. Sensors. 2023;23(15):6887. https://doi.org/10.3390/s23156887

The authors declare that there are no conflicts of interest present.