Artificial intelligence-driven classification method of grapevine major phenological stages using conventional RGB imaging
Accurate monitoring of grapevine phenological stages is essential for optimising vineyard management. This study evaluates the performance of three deep learning architectures (ResNet-34, YOLOv11-Classification and Vision Transformer (ViT)) for automated classification of vineyard canopy images int...
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
International Viticulture and Enology Society
2025-06-01
|
| Series: | OENO One |
| Subjects: | |
| Online Access: | https://oeno-one.eu/article/view/9306 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849422644302053376 |
|---|---|
| author | Ruben Íñiguez Fikile Wolela María Ignacia Gonzalez Pavez Ignacio Barrio Javier Tardáguila Talitha Venter Carlos Poblete-Echeverria |
| author_facet | Ruben Íñiguez Fikile Wolela María Ignacia Gonzalez Pavez Ignacio Barrio Javier Tardáguila Talitha Venter Carlos Poblete-Echeverria |
| author_sort | Ruben Íñiguez |
| collection | DOAJ |
| description |
Accurate monitoring of grapevine phenological stages is essential for optimising vineyard management. This study evaluates the performance of three deep learning architectures (ResNet-34, YOLOv11-Classification and Vision Transformer (ViT)) for automated classification of vineyard canopy images into four key phenological stages: i) Shoot and inflorescence development (E-L 12–18), ii) Flowering (E-L 19–26), iii) Berry formation (E-L 27–33), and iv) Berry ripening (E-L 35–38). These categories correspond to broad developmental periods that may span several E-L stages. A dataset comprising 4,381 images was used to train and validate the models, incorporating data augmentation techniques to improve robustness. Results indicate that all three models achieved high classification accuracy, with ResNet-34 obtaining the highest accuracy (97.4 % validation, 95.6 % test), reinforcing its strong feature extraction capabilities. However, its lower F1-score (95.3 % validation, 91.8 % test) suggests challenges in handling class imbalances. YOLOv11-Classification demonstrated the most balanced classification performance, achieving a high F1-score (93.6 % validation, 91.8 % test) while maintaining the fastest training time, making it particularly suitable for real-time applications. ViT exhibited competitive classification performance but had higher computational demands, limiting its feasibility for real-time vineyard monitoring. A confusion matrix analysis highlighted misclassification trends, particularly between early shoot development and flowering, due to their visual similarities. Despite these challenges, the study confirms that AI models can effectively automate vineyard phenology classification, reducing manual assessment efforts, and contributing to more efficient viticultural decision-making.
|
| format | Article |
| id | doaj-art-eddfe8e59aee4b2a85d4d7f8f047db91 |
| institution | Kabale University |
| issn | 2494-1271 |
| language | English |
| publishDate | 2025-06-01 |
| publisher | International Viticulture and Enology Society |
| record_format | Article |
| series | OENO One |
| spelling | doaj-art-eddfe8e59aee4b2a85d4d7f8f047db912025-08-20T03:30:57ZengInternational Viticulture and Enology SocietyOENO One2494-12712025-06-0159210.20870/oeno-one.2025.59.2.9306Artificial intelligence-driven classification method of grapevine major phenological stages using conventional RGB imagingRuben Íñiguez0Fikile Wolela1María Ignacia Gonzalez Pavez 2Ignacio Barrio3Javier Tardáguila 4Talitha Venter5Carlos Poblete-Echeverria6https://orcid.org/0000-0001-8025-5879Televitis Research Group, University of La Rioja, 26006 Logroño, Spain/Affiliation Institute of Grapevine and Wine Sciences (University of La Rioja, Consejo Superior de Investigaciones Científicas, Gobierno de La Rioja), 26007 Logroño, SpainSouth African Grape and Wine Research Institute (SAGWRI), Stellenbosch University, Private Bag X1, Matieland 7602, South AfricaSouth African Grape and Wine Research Institute (SAGWRI), Stellenbosch University, Private Bag X1, Matieland 7602, South Africa/Research and Extension Center for Irrigation and Agroclimatology (CITRA), Faculty of Agricultural Sciences, Universidad de Talca, Campus Talca, ChileTelevitis Research Group, University of La Rioja, 26006 Logroño, Spain/Affiliation Institute of Grapevine and Wine Sciences (University of La Rioja, Consejo Superior de Investigaciones Científicas, Gobierno de La Rioja), 26007 Logroño, SpainTelevitis Research Group, University of La Rioja, 26006 Logroño, Spain/Affiliation Institute of Grapevine and Wine Sciences (University of La Rioja, Consejo Superior de Investigaciones Científicas, Gobierno de La Rioja), 26007 Logroño, SpainSouth African Grape and Wine Research Institute (SAGWRI), Stellenbosch University, Private Bag X1, Matieland 7602, South AfricaSouth African Grape and Wine Research Institute (SAGWRI), Stellenbosch University, Private Bag X1, Matieland 7602, South Africa Accurate monitoring of grapevine phenological stages is essential for optimising vineyard management. This study evaluates the performance of three deep learning architectures (ResNet-34, YOLOv11-Classification and Vision Transformer (ViT)) for automated classification of vineyard canopy images into four key phenological stages: i) Shoot and inflorescence development (E-L 12–18), ii) Flowering (E-L 19–26), iii) Berry formation (E-L 27–33), and iv) Berry ripening (E-L 35–38). These categories correspond to broad developmental periods that may span several E-L stages. A dataset comprising 4,381 images was used to train and validate the models, incorporating data augmentation techniques to improve robustness. Results indicate that all three models achieved high classification accuracy, with ResNet-34 obtaining the highest accuracy (97.4 % validation, 95.6 % test), reinforcing its strong feature extraction capabilities. However, its lower F1-score (95.3 % validation, 91.8 % test) suggests challenges in handling class imbalances. YOLOv11-Classification demonstrated the most balanced classification performance, achieving a high F1-score (93.6 % validation, 91.8 % test) while maintaining the fastest training time, making it particularly suitable for real-time applications. ViT exhibited competitive classification performance but had higher computational demands, limiting its feasibility for real-time vineyard monitoring. A confusion matrix analysis highlighted misclassification trends, particularly between early shoot development and flowering, due to their visual similarities. Despite these challenges, the study confirms that AI models can effectively automate vineyard phenology classification, reducing manual assessment efforts, and contributing to more efficient viticultural decision-making. https://oeno-one.eu/article/view/9306deep learningvineyard phenologyYOLOv11ResNet-34Vision Transformer (ViT)precision viticulture |
| spellingShingle | Ruben Íñiguez Fikile Wolela María Ignacia Gonzalez Pavez Ignacio Barrio Javier Tardáguila Talitha Venter Carlos Poblete-Echeverria Artificial intelligence-driven classification method of grapevine major phenological stages using conventional RGB imaging OENO One deep learning vineyard phenology YOLOv11 ResNet-34 Vision Transformer (ViT) precision viticulture |
| title | Artificial intelligence-driven classification method of grapevine major phenological stages using conventional RGB imaging |
| title_full | Artificial intelligence-driven classification method of grapevine major phenological stages using conventional RGB imaging |
| title_fullStr | Artificial intelligence-driven classification method of grapevine major phenological stages using conventional RGB imaging |
| title_full_unstemmed | Artificial intelligence-driven classification method of grapevine major phenological stages using conventional RGB imaging |
| title_short | Artificial intelligence-driven classification method of grapevine major phenological stages using conventional RGB imaging |
| title_sort | artificial intelligence driven classification method of grapevine major phenological stages using conventional rgb imaging |
| topic | deep learning vineyard phenology YOLOv11 ResNet-34 Vision Transformer (ViT) precision viticulture |
| url | https://oeno-one.eu/article/view/9306 |
| work_keys_str_mv | AT rubeniniguez artificialintelligencedrivenclassificationmethodofgrapevinemajorphenologicalstagesusingconventionalrgbimaging AT fikilewolela artificialintelligencedrivenclassificationmethodofgrapevinemajorphenologicalstagesusingconventionalrgbimaging AT mariaignaciagonzalezpavez artificialintelligencedrivenclassificationmethodofgrapevinemajorphenologicalstagesusingconventionalrgbimaging AT ignaciobarrio artificialintelligencedrivenclassificationmethodofgrapevinemajorphenologicalstagesusingconventionalrgbimaging AT javiertardaguila artificialintelligencedrivenclassificationmethodofgrapevinemajorphenologicalstagesusingconventionalrgbimaging AT talithaventer artificialintelligencedrivenclassificationmethodofgrapevinemajorphenologicalstagesusingconventionalrgbimaging AT carlospobleteecheverria artificialintelligencedrivenclassificationmethodofgrapevinemajorphenologicalstagesusingconventionalrgbimaging |