Fang Jiangxiong, Liu Huaxiang, Zhang Shiqing, Hu Hui, Gu Huaqi, Fu Youyao
Institute of Intelligent Information Processing, Taizhou University, Taizhou, Zhejiang, China.
Department of Information and Remote Sensing, Jiangxi Provincial Natural Resources Development Center, Nanchang, Jiangxi, China.
Front Plant Sci. 2025 Apr 8;16:1515105. doi: 10.3389/fpls.2025.1515105. eCollection 2025.
Pepper leaf segmentation plays a pivotal role in monitoring pepper leaf diseases across diverse backgrounds and ensuring healthy pepper growth. However, existing Transformer-based segmentation methods grapple with computational inefficiency, excessive parameterization, and inadequate utilization of edge information.
To address these challenges, this study introduces an Adaptive Multi-Scale MLP (AMS-MLP) framework. This framework integrates the Multi-Path Aggregation Module (MPAM) and the Multi-Scale Context Relation Mask Module (MCRD) to refine object boundaries in pepper leaf segmentation. The AMS-MLP includes an encoder, an Adaptive Multi-Scale MLP (AM-MLP) module, and a decoder. The encoder's MPAM fuses five-scale features for accurate boundary extraction. The AM-MLP splits features into global and local branches, with an adaptive attention mechanism balancing them. The decoder enhances boundary feature extraction using MCRD.
To validate the proposed method, extensive experiments were conducted on three pepper leaf datasets with varying backgrounds. Results demonstrate mean Intersection over Union (mIoU) scores of 97.39%, 96.91%, and 97.91%, and F1 scores of 98.29%, 97.86%, and 98.51% across the datasets, respectively.
Comparative analysis with U-Net and state-of-the-art models reveals that the proposed method significantly improves the accuracy and efficiency of pepper leaf image segmentation.
辣椒叶片分割在监测不同背景下的辣椒叶病害以及确保辣椒健康生长方面发挥着关键作用。然而，现有的基于Transformer的分割方法存在计算效率低、参数化过多以及边缘信息利用不足等问题。
为应对这些挑战，本研究引入了一种自适应多尺度MLP（AMS-MLP）框架。该框架集成了多路径聚合模块（MPAM）和多尺度上下文关系掩码模块（MCRD），以优化辣椒叶分割中的对象边界。AMS-MLP包括一个编码器、一个自适应多尺度MLP（AM-MLP）模块和一个解码器。编码器的MPAM融合五尺度特征以进行精确的边界提取。AM-MLP将特征分为全局和局部分支，并通过自适应注意力机制对它们进行平衡。解码器使用MCRD增强边界特征提取。
为验证所提出的方法，在三个具有不同背景的辣椒叶数据集上进行了广泛的实验。结果表明，在各个数据集上，平均交并比（mIoU）分数分别为97.39%、96.91%和97.91%，F1分数分别为98.29%、97.86%和98.51%。
与U-Net和现有最先进模型的对比分析表明，所提出的方法显著提高了辣椒叶图像分割的准确性和效率。
