番茄成熟度视觉判定新框架刊于《科学报告》：约束分类决定落在果实区域内、多骨干网络实验均有提升
《科学报告》11月14日发表一项研究，提出一种名为「区域一致性约束学习」的框架，用于按外观图像判定番茄成熟度。作者来自东北电力大学，团队成员分布在该校计算机学院、图书馆与电气工程学院，通讯作者为楼建勋。
在农业生产的质量控制中，按外观准确判定番茄成熟度是常见需求，但在背景杂乱、成熟特征不明确的复杂视觉条件下并不容易。作者指出，已有方法能在一定程度上抑制背景干扰，却缺少明确机制保证分类决定形成于与成熟度相关的果实区域之内。
新框架把两类信息结合起来：一类是由类激活图刻画的、与分类决定相关的区域，另一类是基于显著性的空间先验。训练过程中，框架强制这两类区域保持一致，从而约束分类决定形成的位置；这一过程不修改输入图像，也不改动中间特征表示。
作者在一个真实场景的番茄数据集上做了实验。结果显示，该框架在多种骨干网络架构上都能持续提升分类表现，取得的提升具有统计显著性，并且在推理阶段不增加额外计算成本。
作者表示，该框架为复杂农业条件下提高视觉成熟度判定的可靠性提供了一种有效途径。论文2026年1月14日投稿，10月12日被接受，11月14日发表，数字对象标识符为 10.1038/s41598-026-76301-6。
论文以知识共享署名4.0国际许可开放获取，作者声明无利益冲突。
图中是藤上的番茄植株，果实多尚未成熟、个别开始转红
New Framework for Visual Tomato Maturity Assessment in Scientific Reports: Keeping Classification Decisions Inside the Fruit Region
A study published in Scientific Reports on November 14 proposes a region-consistency constrained learning framework for assessing tomato maturity from visual appearance. The authors are based at Northeast Electric Power University in Jilin, China, across its School of Computer Science, its Library, and its School of Electrical Engineering, with Jianxun Lou as corresponding author.
Accurate tomato maturity assessment based on visual appearance is essential for quality control in agricultural production, yet remains challenging under complex visual conditions due to background clutter and ambiguous maturity cues. The authors note that many existing methods can suppress background interference to some extent, but lack explicit mechanisms to ensure that classification decisions are formed within maturity-relevant fruit regions.
The framework integrates decision-relevant regions characterized by Class Activation Maps (CAM) with saliency-based spatial priors and enforces consistency between them during training, thereby regulating where classification decisions are formed without modifying the input or intermediate feature representations.
Experiments conducted on a real-world tomato dataset demonstrate that the proposed framework consistently improves classification performance across multiple backbone architectures. Statistically significant gains are achieved without introducing additional computational cost during inference.
The authors conclude that the framework provides an effective way to enhance the reliability of visual maturity assessment under complex agricultural conditions. The article was received on January 14, 2026, accepted on October 12, 2026, and published on November 14, 2026, with the digital object identifier 10.1038/s41598-026-76301-6.
The article is published open access under a Creative Commons Attribution 4.0 International License, and the authors declare no competing interests.
来源：《科学报告》
发布：GSM·美国（发布链接）
图片来源：图：藤上的番茄植株，果实多尚未成熟、个别开始转红。来源：NPS, Public domain，维基共享资源。 / Photo: Tomato plants on the vine, mostly unripe with a few fruits starting to turn. Credit: NPS, Public domain, via Wikimedia Commons.
GSM Global · 本篇英文版 English edition of this story
