Unverified paper record
Maturity and size estimation with yield mapping for hydroponic strawberries using machine vision
Smart Agricultural Technology · 20 Jul 2026 · 10.1016/j.atech.2026.102416
Abstract
Strawberry production in hydroponic systems requires precise and spatially explicit information on fruit load, maturity, and size to support harvest planning and quality control. This study presents an integrated computer vision and GNSS RTK framework that detects individual strawberries, estimates their maturity and physical dimensions, assigns fruits to operational size–maturity categories, and generates high-resolution yield maps in an open-field hydroponic system. RGB-D images are processed with YOLOv8-seg and a BoT-SORT-based tracking module to obtain instance-level masks, from which a color-based ripeness index and 3D point clouds are derived. After DBSCAN-based filtering, fruit length and width are computed in metric space and combined with ripeness percentage to assign each strawberry to one of nine operational size–maturity categories using predefined threshold rules. The system was evaluated in a commercial hydroponic crop in Arcabuco, Boyacá, Colombia, achieving accurate segmentation, with median IoU values up to 0.83 for bounding boxes and 0.71 for masks, and mean absolute errors of 2.51 mm in length and 1.85 mm in width with respect to Vernier caliper measurements. Yield maps aggregated in 1 m2 cells and by crop row revealed marked spatial variability in fruit density, maturity state, and size–maturity composition, allowing the identification of zones with a high concentration of fruits ready for harvest versus areas dominated by immature fruits. The proposed framework provides a practical and low-cost decision support tool for hydroponic strawberry management and may be adapted to other high-value horticultural crops after recalibration under different crop architectures, cultivars, and environmental conditions.
Plant phenotyping relevance
RGB-D画像と追跡・3D解析により、イチゴ果実の成熟度、寸法、果実密度および収穫状態を個体レベルで推定し、精度検証も行う中心的なフェノタイピング手法研究である。
abstractThis study presents an integrated computer vision and GNSS RTK framework that detects individual strawberries, estimates their maturity and physical dimensions, assigns fruits to operational size–maturity categories, and generates high-resolution yield maps in an open-field hydroponic system.
abstractThe system was evaluated in a commercial hydroponic crop in Arcabuco, Boyacá, Colombia, achieving accurate segmentation, with median IoU values up to 0.83 for bounding boxes and 0.71 for masks, and mean absolute errors of 2.51 mm in length and 1.85 mm in width with respect to Vernier caliper measurements.
Code and data availability
公開論文であることは確認できましたが、現在の公式API・許可済み取得経路では本文を自動取得できませんでした。
No evidence-backed public reproduction asset is currently recorded.
This is an automatically classified, unverified record. Curator approval is required before any resource enters the Catalog.