Multi-objective RGB-D fusion network for non-destructive strawberry trait assessment

Growing consumer demand for high-quality strawberries has highlighted the need for accurate, efficient, and non-destructive methods to assess key postharvest quality traits, such as weight, size uniformity, and quantity. This study proposes a multi-objective learning algorithm that leverages RGB-D m...

Full description

Saved in:
Bibliographic Details
Main Authors: Zhenzhen Cheng, Yifan Cheng, Bailing Miao, Tingting Fang, Shoufu Gong
Format: Article
Language:English
Published: Frontiers Media S.A. 2025-03-01
Series:Frontiers in Plant Science
Subjects:
Online Access:https://www.frontiersin.org/articles/10.3389/fpls.2025.1564301/full
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850069075069239296
author Zhenzhen Cheng
Yifan Cheng
Bailing Miao
Tingting Fang
Shoufu Gong
author_facet Zhenzhen Cheng
Yifan Cheng
Bailing Miao
Tingting Fang
Shoufu Gong
author_sort Zhenzhen Cheng
collection DOAJ
description Growing consumer demand for high-quality strawberries has highlighted the need for accurate, efficient, and non-destructive methods to assess key postharvest quality traits, such as weight, size uniformity, and quantity. This study proposes a multi-objective learning algorithm that leverages RGB-D multimodal information to estimate these quality metrics. The algorithm develops a fusion expert network architecture that maximizes the use of multimodal features while preserving the distinct details of each modality. Additionally, a novel Heritable Loss function is implemented to reduce redundancy and enhance model performance. Experimental results show that the coefficient of determination (R²) values ​​for weight, size uniformity and number are 0.94, 0.90 and 0.95 respectively. Ablation studies demonstrate the advantage of the architecture in multimodal, multi-task prediction accuracy. Compared to single-modality models, non-fusion branch networks, and attention-enhanced fusion models, our approach achieves enhanced performance across multi-task learning scenarios, providing more precise data for trait assessment and precision strawberry applications.
format Article
id doaj-art-eb77999a2fcc48868f57071a1e2249c4
institution DOAJ
issn 1664-462X
language English
publishDate 2025-03-01
publisher Frontiers Media S.A.
record_format Article
series Frontiers in Plant Science
spelling doaj-art-eb77999a2fcc48868f57071a1e2249c42025-08-20T02:47:52ZengFrontiers Media S.A.Frontiers in Plant Science1664-462X2025-03-011610.3389/fpls.2025.15643011564301Multi-objective RGB-D fusion network for non-destructive strawberry trait assessmentZhenzhen Cheng0Yifan Cheng1Bailing Miao2Tingting Fang3Shoufu Gong4Department of Horticulture, Xinyang Agriculture and Forestry University, Xinyang, ChinaDepartment of Optical and Electronic Information, Huazhong University of Science and Technology, Wuhan, ChinaDepartment of Horticulture, Xinyang Agriculture and Forestry University, Xinyang, ChinaDepartment of Horticulture, Xinyang Agriculture and Forestry University, Xinyang, ChinaDepartment of Horticulture, Xinyang Agriculture and Forestry University, Xinyang, ChinaGrowing consumer demand for high-quality strawberries has highlighted the need for accurate, efficient, and non-destructive methods to assess key postharvest quality traits, such as weight, size uniformity, and quantity. This study proposes a multi-objective learning algorithm that leverages RGB-D multimodal information to estimate these quality metrics. The algorithm develops a fusion expert network architecture that maximizes the use of multimodal features while preserving the distinct details of each modality. Additionally, a novel Heritable Loss function is implemented to reduce redundancy and enhance model performance. Experimental results show that the coefficient of determination (R²) values ​​for weight, size uniformity and number are 0.94, 0.90 and 0.95 respectively. Ablation studies demonstrate the advantage of the architecture in multimodal, multi-task prediction accuracy. Compared to single-modality models, non-fusion branch networks, and attention-enhanced fusion models, our approach achieves enhanced performance across multi-task learning scenarios, providing more precise data for trait assessment and precision strawberry applications.https://www.frontiersin.org/articles/10.3389/fpls.2025.1564301/fullstrawberry qualityfruit traits estimationcomputer visiondeep learningRGB-D modality fusion
spellingShingle Zhenzhen Cheng
Yifan Cheng
Bailing Miao
Tingting Fang
Shoufu Gong
Multi-objective RGB-D fusion network for non-destructive strawberry trait assessment
Frontiers in Plant Science
strawberry quality
fruit traits estimation
computer vision
deep learning
RGB-D modality fusion
title Multi-objective RGB-D fusion network for non-destructive strawberry trait assessment
title_full Multi-objective RGB-D fusion network for non-destructive strawberry trait assessment
title_fullStr Multi-objective RGB-D fusion network for non-destructive strawberry trait assessment
title_full_unstemmed Multi-objective RGB-D fusion network for non-destructive strawberry trait assessment
title_short Multi-objective RGB-D fusion network for non-destructive strawberry trait assessment
title_sort multi objective rgb d fusion network for non destructive strawberry trait assessment
topic strawberry quality
fruit traits estimation
computer vision
deep learning
RGB-D modality fusion
url https://www.frontiersin.org/articles/10.3389/fpls.2025.1564301/full
work_keys_str_mv AT zhenzhencheng multiobjectivergbdfusionnetworkfornondestructivestrawberrytraitassessment
AT yifancheng multiobjectivergbdfusionnetworkfornondestructivestrawberrytraitassessment
AT bailingmiao multiobjectivergbdfusionnetworkfornondestructivestrawberrytraitassessment
AT tingtingfang multiobjectivergbdfusionnetworkfornondestructivestrawberrytraitassessment
AT shoufugong multiobjectivergbdfusionnetworkfornondestructivestrawberrytraitassessment