Use of YOLOv5 Trained Model for Robotic Courgette Harvesting and Efficiency Analysis

Erhan Kahya

Makale detayları

Use of YOLOv5 Trained Model for Robotic Courgette Harvesting and Efficiency Analysis

Yazarlar:

Erhan Kahya

Yayıncı:

Van Yuzuncu Yıl University

Ana Başlıklar:

Agricultural Automatization

Dergi:

Yuzuncu Yıl University Journal of Agricultural Sciences Vol.34 Issue 4, pp.669-689

Belge Tipi:

Makale

Yayın Tarihi:

12/2024

Belge Dili:

İngilizce

Document ID (Org No.):

2001333263

pISSN:

1308-7576

Anahtar kelimeler:

Özet:

 The utilization of machine learning in vegetable harvesting not only enhances efficiency and precision but also addresses labor shortages and improves overall agricultural productivity. In this study, a machine learning method was developed for harvesting courgette fruit. Courgette is a fruit that can take a long time to select and harvest in the agricultural area where it is grown. The YOLOv5 models nano, small, medium, and large were used as a deep learning method. All metric values of the models were analyzed. The most successful model was the one trained with the YOLOv5m algorithm using 20 batches and 160 epochs with 640x640 images. The results of the model scores were analyzed as metricsprecision, metricsrecall, metricsmAP_0.5 and metricsmAP_0.5: 0.95. These metrics are key indicators that measure the recognition success of a model and reflect the performance of the respective model on the validation dataset. The metrics data of the YOLOv5 medium model proved to be higher compared to the other models. The measured values were YOLOv5m  size: 640x640, batch: 20, epoch: 160, algorithm: YOLOv5m. It was concluded that YOLOv5m is the best recognition model that can be used in robotic courgette harvesting to separate the courgette from the branch.

 The utilization of machine learning in vegetable harvesting not only enhances efficiency and precision but also addresses labor shortages and improves overall agricultural productivity. In this study, a machine learning method was developed for harvesting courgette fruit. Courgette is a fruit that can take a long time to select and harvest in the agricultural area where it is grown. The YOLOv5 models nano, small, medium, and large were used as a deep learning method. All metric values of the models were analyzed. The most successful model was the one trained with the YOLOv5m algorithm using 20 batches and 160 epochs with 640x640 images. The results of the model scores were analyzed as metricsprecision, metricsrecall, metricsmAP_0.5 and metricsmAP_0.5: 0.95. These metrics are key indicators that measure the recognition success of a model and reflect the performance of the respective model on the validation dataset. The metrics data of the YOLOv5 medium model proved to be higher compared to the other models. The measured values were YOLOv5m  size: 640x640, batch: 20, epoch: 160, algorithm: YOLOv5m. It was concluded that YOLOv5m is the best recognition model that can be used in robotic courgette harvesting to separate the courgette from the branch.

 The utilization of machine learning in vegetable harvesting not only enhances efficiency and precision but also addresses labor shortages and improves overall agricultural productivity. In this study, a machine learning method was developed for harvesting courgette fruit. Courgette is a fruit that can take a long time to select and harvest in the agricultural area where it is grown. The YOLOv5 models nano, small, medium, and large were used as a deep learning method. All metric values of the models were analyzed. The most successful model was the one trained with the YOLOv5m algorithm using 20 batches and 160 epochs with 640x640 images. The results of the model scores were analyzed as metricsprecision, metricsrecall, metricsmAP_0.5 and metricsmAP_0.5: 0.95. These metrics are key indicators that measure the recognition success of a model and reflect the performance of the respective model on the validation dataset. The metrics data of the YOLOv5 medium model proved to be higher compared to the other models. The measured values were YOLOv5m  size: 640x640, batch: 20, epoch: 160, algorithm: YOLOv5m. It was concluded that YOLOv5m is the best recognition model that can be used in robotic courgette harvesting to separate the courgette from the branch.

 The utilization of machine learning in vegetable harvesting not only enhances efficiency and precision but also addresses labor shortages and improves overall agricultural productivity. In this study, a machine learning method was developed for harvesting courgette fruit. Courgette is a fruit that can take a long time to select and harvest in the agricultural area where it is grown. The YOLOv5 models nano, small, medium, and large were used as a deep learning method. All metric values of the models were analyzed. The most successful model was the one trained with the YOLOv5m algorithm using 20 batches and 160 epochs with 640x640 images. The results of the model scores were analyzed as metricsprecision, metricsrecall, metricsmAP_0.5 and metricsmAP_0.5: 0.95. These metrics are key indicators that measure the recognition success of a model and reflect the performance of the respective model on the validation dataset. The metrics data of the YOLOv5 medium model proved to be higher compared to the other models. The measured values were YOLOv5m  size: 640x640, batch: 20, epoch: 160, algorithm: YOLOv5m. It was concluded that YOLOv5m is the best recognition model that can be used in robotic courgette harvesting to separate the courgette from the branch.

Makale detayları

Use of YOLOv5 Trained Model for Robotic Courgette Harvesting and Efficiency Analysis

Yazarlar:

Erhan Kahya

Yayıncı:

Van Yuzuncu Yıl University

Ana Başlıklar:

Agricultural Automatization

Dergi:

Yuzuncu Yıl University Journal of Agricultural Sciences Vol.34 Issue 4, pp.669-689

Belge Tipi:

Makale

Yayın Tarihi:

12/2024

Belge Dili:

İngilizce

Document ID (Org No.):

2001333263

pISSN:

1308-7576

deep learning

courgette

YOLOv5

Product forecasting

Sayı

Makale detayları

Use of YOLOv5 Trained Model for Robotic Courgette Harvesting and Efficiency Analysis

Yazarlar: Erhan Kahya

Yayıncı: Van Yuzuncu Yıl University

Ana Başlıklar: Agricultural Automatization

Dergi: Yuzuncu Yıl University Journal of Agricultural Sciences Vol.34 Issue 4, pp.669-689

Belge Tipi: Makale

Yayın Tarihi: 12/2024

Belge Dili: İngilizce

Document ID (Org No.): 2001333263

pISSN: 1308-7576

deep learning

courgette

YOLOv5

Product forecasting

Sayı

Yazarlar:

Erhan Kahya

Yayıncı:

Van Yuzuncu Yıl University

Ana Başlıklar:

Agricultural Automatization

Dergi:

Yuzuncu Yıl University Journal of Agricultural Sciences Vol.34 Issue 4, pp.669-689

Belge Tipi:

Makale

Yayın Tarihi:

12/2024

Belge Dili:

İngilizce

Document ID (Org No.):

2001333263

pISSN:

1308-7576