Scholarlink

<p>Manual maturity assessment in greenhouses is labor-intensive and increasingly constrained by labor shortages, limiting timely harvest and quality control. To address this challenge, this study proposes an original approach for real-time monitoring the growth and maturity degree of mini tomatoes in complex greenhouse environments. The proposed framework combines visual features by YOLOv8-based segmentation and skeleton and environmental features for maturity estimation. This approach also employs CIELAB color space and HUE angle with an <InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(L*\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mi>L</mi> <mrow /> <mo>∗</mo> </mrow> </math></EquationSource> </InlineEquation> threshold filter against the variable lightness situation and occlusion in the greenhouse. The segmentation and skeleton models achieved reliable performance of mask mAP of 80.3% and skeleton precision of 97%. This enables the accurate mini tomato fruit detection and cluster structure analysis. The employed <InlineEquation ID="IEq2"> <EquationSource Format="TEX">\(L*\)</EquationSource> <EquationSource Format="MATHML"><math> <mrow> <mi>L</mi> <mrow /> <mo>∗</mo> </mrow> </math></EquationSource> </InlineEquation> filter improved color level estimation accuracy from 83 to 95%. Especially, the harvest-ready stage increases from 64 to 94%. Additionally, a linear regression model confirmed the relationship between HUE angle and cumulative temperature. These results show that the proposed system can support efficient, data-driven monitoring of mini tomato growth stages.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Maturity estimation of mini tomatoes with color analysis and cumulative temperature

  • 期刊论文

  • 发表日期：

    2025年12月16日

查看全文
  • Yan Lyu,
  • Kyuki Shibuya,
  • Yoshiki Gama,
  • Yoshiki Hatanaka,
  • Warut Timprae,
  • Poltak Sandro Rumahorbo,
  • Shinya Yamanaka,
  • Shinya Watanabe

摘要

Manual maturity assessment in greenhouses is labor-intensive and increasingly constrained by labor shortages, limiting timely harvest and quality control. To address this challenge, this study proposes an original approach for real-time monitoring the growth and maturity degree of mini tomatoes in complex greenhouse environments. The proposed framework combines visual features by YOLOv8-based segmentation and skeleton and environmental features for maturity estimation. This approach also employs CIELAB color space and HUE angle with an \(L*\) L ∗ threshold filter against the variable lightness situation and occlusion in the greenhouse. The segmentation and skeleton models achieved reliable performance of mask mAP of 80.3% and skeleton precision of 97%. This enables the accurate mini tomato fruit detection and cluster structure analysis. The employed \(L*\) L ∗ filter improved color level estimation accuracy from 83 to 95%. Especially, the harvest-ready stage increases from 64 to 94%. Additionally, a linear regression model confirmed the relationship between HUE angle and cumulative temperature. These results show that the proposed system can support efficient, data-driven monitoring of mini tomato growth stages.