Sulfur-Fumigated Ginger Identification Method Based on Meta-Learning for Different Devices DOI Creative Commons
Tianshu Wang, Jiawang He, Hui Yan

и другие.

Foods, Год журнала: 2024, Номер 13(23), С. 3870 - 3870

Опубликована: Ноя. 29, 2024

Since ginger has characteristics of both food and medicine, it a significant market demand worldwide. To effectively store achieve the drying color enhancement effects required for better sales, is often subjected to sulfur fumigation. Although fumigation methods can prevent from becoming moldy, they cause residual dioxide, harming human health. Traditional detection face disadvantages such as complex operation, high time consumption, easy consumption. This paper presents sulfur-fumigated method based on natural image recognition. By directly using images mobile phones, proposed achieves non-destructive testing reduces operational complexity. First, four phones different brands are used collect sulfur- non-sulfur-fumigated samples. Then, preprocessed remove blank background in deep neural network designed extract features images. Next, recognition model generated features. Finally, meta-learning parameters introduced enable learn adapt new tasks, thereby improving adaptability model. Thus, devices its real application. The experimental results indicate that recall rate, F1 score, AUC-ROC more than 0.9, discrimination accuracy these above 0.95. Therefore, this good predictive ability excellent practical value identifying ginger.

Язык: Английский

Quality Evaluation of Wenyujin Rhizoma Concisum From Different Districts in China Based on HPLC, Heracles NEO Ultrafast Gas‐Phase Electronic Nose, and FT‐NIR DOI Creative Commons

S J Wang,

Kewei Zhang,

Xiuqi Gan

и другие.

Phytochemical Analysis, Год журнала: 2025, Номер unknown

Опубликована: Янв. 12, 2025

ABSTRACT Introduction Wenyujin Rhizoma Concisum, named as Pian Jianghuang (PJH) in China, is the decoction piece from dried rhizome of Curcuma wenyujin Y. H. Chenet C. Ling, has been used to relieve pain for many years China. However, qualities PJH different districts differ greatly due their growing environments, which would affect clinical applications. Objective To evaluate Methods HPLC, Heracles NEO ultrafast gas‐phase electronic nose, and Fourier transform near‐infrared (FT‐NIR) spectroscopy were applied estimate Results By average contents neocurdione, curdione, germacrone, furanodiene 0.203%, 0.151%, 0.022%, 0.022% Fujian (FJ); 0.447%, 0.786%, 0.298%, 0.276% Zhejiang (ZJ); 0.082%, 0.259%, 0.038%, 0.019% Yunnan (YN); 0.041%, 0.260%, 0.024% Guangxi (GX); 0.026%, 0.091%, 0.016% Anhui (AH), respectively. unique odor components FJ, YN, GX 1,2,4‐thiadiazole,5‐ethoxy‐3‐(trichloromethyl), 5,6,7,8‐tetrahydroquinoxaline, 1,3,5‐trimethylbenzene, respectively, while ZJ AH both contained two components, respectively: pentyl pentanoate dill ether (ZJ), myrcene, 2‐pentadecanone (AH). Moreover, above five could be discerned quickly by FT‐NIR. Conclusion The application multidimensional analytical techniques quality assessment China provide a new idea control geographical origin traceability traditional Chinese materia medica.

Язык: Английский

Процитировано

0
Sulfur-Fumigated Ginger Identification Method Based on Meta-Learning for Different Devices DOI Creative Commons
Tianshu Wang, Jiawang He, Hui Yan

и другие.

Foods, Год журнала: 2024, Номер 13(23), С. 3870 - 3870

Опубликована: Ноя. 29, 2024

Since ginger has characteristics of both food and medicine, it a significant market demand worldwide. To effectively store achieve the drying color enhancement effects required for better sales, is often subjected to sulfur fumigation. Although fumigation methods can prevent from becoming moldy, they cause residual dioxide, harming human health. Traditional detection face disadvantages such as complex operation, high time consumption, easy consumption. This paper presents sulfur-fumigated method based on natural image recognition. By directly using images mobile phones, proposed achieves non-destructive testing reduces operational complexity. First, four phones different brands are used collect sulfur- non-sulfur-fumigated samples. Then, preprocessed remove blank background in deep neural network designed extract features images. Next, recognition model generated features. Finally, meta-learning parameters introduced enable learn adapt new tasks, thereby improving adaptability model. Thus, devices its real application. The experimental results indicate that recall rate, F1 score, AUC-ROC more than 0.9, discrimination accuracy these above 0.95. Therefore, this good predictive ability excellent practical value identifying ginger.

Язык: Английский

Процитировано

0