Identification of Individual <i>Hanwoo</i> Cattle by Muzzle Pattern Images through Deep Learning
The objective of this study was to identify <i>Hanwoo</i> cattle via a deep-learning model using muzzle images. A total of 9230 images from 336 <i>Hanwoo</i> were used. Images of the same individuals were taken at four different times to avoid overfitted models. Muzzle images...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2023-09-01
|
| Series: | Animals |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2076-2615/13/18/2856 |
| _version_ | 1797581609598189568 |
|---|---|
| author | Taejun Lee Youngjun Na Beob Gyun Kim Sangrak Lee Yongjun Choi |
| author_facet | Taejun Lee Youngjun Na Beob Gyun Kim Sangrak Lee Yongjun Choi |
| author_sort | Taejun Lee |
| collection | DOAJ |
| description | The objective of this study was to identify <i>Hanwoo</i> cattle via a deep-learning model using muzzle images. A total of 9230 images from 336 <i>Hanwoo</i> were used. Images of the same individuals were taken at four different times to avoid overfitted models. Muzzle images were cropped by the YOLO v8-based model trained with 150 images with manual annotation. Data blocks were composed of image and national livestock traceability numbers and were randomly selected and stored as train, validation test data. Transfer learning was performed with the tiny, small and medium versions of Efficientnet v2 models with SGD, RMSProp, Adam and Lion optimizers. The small version using Lion showed the best validation accuracy of 0.981 in 36 epochs within 12 transfer-learned models. The top five models achieved the best validation accuracy and were evaluated with the training data for practical usage. The small version using Adam showed the best test accuracy of 0.970, but the small version using RMSProp showed the lowest repeated error. Results with high accuracy prediction in this study demonstrated the potential of muzzle patterns as an identification key for individual cattle. |
| first_indexed | 2024-03-10T23:07:50Z |
| format | Article |
| id | doaj.art-bb0c5c43854f47f583ea446142e14b4e |
| institution | Directory Open Access Journal |
| issn | 2076-2615 |
| language | English |
| last_indexed | 2024-03-10T23:07:50Z |
| publishDate | 2023-09-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Animals |
| spelling | doaj.art-bb0c5c43854f47f583ea446142e14b4e2023-11-19T09:14:33ZengMDPI AGAnimals2076-26152023-09-011318285610.3390/ani13182856Identification of Individual <i>Hanwoo</i> Cattle by Muzzle Pattern Images through Deep LearningTaejun Lee0Youngjun Na1Beob Gyun Kim2Sangrak Lee3Yongjun Choi4Department of Animal Science, Konkuk University, Seoul 05029, Republic of KoreaDepartment of Animal Science, Konkuk University, Seoul 05029, Republic of KoreaDepartment of Animal Science, Konkuk University, Seoul 05029, Republic of KoreaDepartment of Animal Science, Konkuk University, Seoul 05029, Republic of KoreaDepartment of Animal Science, Konkuk University, Seoul 05029, Republic of KoreaThe objective of this study was to identify <i>Hanwoo</i> cattle via a deep-learning model using muzzle images. A total of 9230 images from 336 <i>Hanwoo</i> were used. Images of the same individuals were taken at four different times to avoid overfitted models. Muzzle images were cropped by the YOLO v8-based model trained with 150 images with manual annotation. Data blocks were composed of image and national livestock traceability numbers and were randomly selected and stored as train, validation test data. Transfer learning was performed with the tiny, small and medium versions of Efficientnet v2 models with SGD, RMSProp, Adam and Lion optimizers. The small version using Lion showed the best validation accuracy of 0.981 in 36 epochs within 12 transfer-learned models. The top five models achieved the best validation accuracy and were evaluated with the training data for practical usage. The small version using Adam showed the best test accuracy of 0.970, but the small version using RMSProp showed the lowest repeated error. Results with high accuracy prediction in this study demonstrated the potential of muzzle patterns as an identification key for individual cattle.https://www.mdpi.com/2076-2615/13/18/2856cattle identificationdeep learningtransfer learningEfficientnet<i>Hanwoo</i>muzzle pattern |
| spellingShingle | Taejun Lee Youngjun Na Beob Gyun Kim Sangrak Lee Yongjun Choi Identification of Individual <i>Hanwoo</i> Cattle by Muzzle Pattern Images through Deep Learning Animals cattle identification deep learning transfer learning Efficientnet <i>Hanwoo</i> muzzle pattern |
| title | Identification of Individual <i>Hanwoo</i> Cattle by Muzzle Pattern Images through Deep Learning |
| title_full | Identification of Individual <i>Hanwoo</i> Cattle by Muzzle Pattern Images through Deep Learning |
| title_fullStr | Identification of Individual <i>Hanwoo</i> Cattle by Muzzle Pattern Images through Deep Learning |
| title_full_unstemmed | Identification of Individual <i>Hanwoo</i> Cattle by Muzzle Pattern Images through Deep Learning |
| title_short | Identification of Individual <i>Hanwoo</i> Cattle by Muzzle Pattern Images through Deep Learning |
| title_sort | identification of individual i hanwoo i cattle by muzzle pattern images through deep learning |
| topic | cattle identification deep learning transfer learning Efficientnet <i>Hanwoo</i> muzzle pattern |
| url | https://www.mdpi.com/2076-2615/13/18/2856 |
| work_keys_str_mv | AT taejunlee identificationofindividualihanwooicattlebymuzzlepatternimagesthroughdeeplearning AT youngjunna identificationofindividualihanwooicattlebymuzzlepatternimagesthroughdeeplearning AT beobgyunkim identificationofindividualihanwooicattlebymuzzlepatternimagesthroughdeeplearning AT sangraklee identificationofindividualihanwooicattlebymuzzlepatternimagesthroughdeeplearning AT yongjunchoi identificationofindividualihanwooicattlebymuzzlepatternimagesthroughdeeplearning |