Scholarlink

<p>Tea is known as an important crop in many parts of South and Southeast Asia, yet the production of tea is still hampered by the multiple diseases that decrease the quantity and quality. Traditional methods of inspection, which are manual, are not consistent, labor-intensive, and depend on extensive monitoring. This paper introduces a lightweight convolutional neural network (CNN) designed for weakly supervised classification under a multi-label framework, trained on single-label image-level annotations and disease localization in tea leaves called LightTeaNet. LightTeaNet learns directly from image-level labels and employs Class Activation Mapping (CAM) to localize disease-affected regions automatically, unlike conventional object detection models such as YOLO, which require extensive bounding box annotations. Importantly, because detector architectures simultaneously optimize for classification, bounding box regression, and objectness estimation, their classification metrics are not directly comparable to those of LightTeaNet, which is optimized solely for image-level classification. For Parameter efficiency, the network integrates Depthwise Separable Convolutions, and for enhanced feature discrimination, it integrates Channel Attention. LightTeaNet has achieved a Precision of 0.9615, a Recall of 0.8772, and an F1-score of 0.9179, while it shows mAP@0.50=0.1810 without any manual annotations, reflecting the inherent coarseness of CAM-based localization under weak supervision. With only 2.32 M parameters, 6.11 GFLOPs, and a GPU inference latency of 2.44 ms (410.40 FPS), these results validate the model as an interpretable and computationally lightweight framework for intelligent disease monitoring in agriculture.</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

LightTeaNet: a weakly supervised lightweight CNN for multi-label tea leaf disease detection and localization

  • 期刊论文

  • 发表日期：

    2026年8月4日

查看全文
  • Naif Haider Chowdhury,
  • Md Rahim,
  • Syed Farhan Hasan,
  • Murad Hasan,
  • Prithwiraj Bhattacharjee

摘要

Tea is known as an important crop in many parts of South and Southeast Asia, yet the production of tea is still hampered by the multiple diseases that decrease the quantity and quality. Traditional methods of inspection, which are manual, are not consistent, labor-intensive, and depend on extensive monitoring. This paper introduces a lightweight convolutional neural network (CNN) designed for weakly supervised classification under a multi-label framework, trained on single-label image-level annotations and disease localization in tea leaves called LightTeaNet. LightTeaNet learns directly from image-level labels and employs Class Activation Mapping (CAM) to localize disease-affected regions automatically, unlike conventional object detection models such as YOLO, which require extensive bounding box annotations. Importantly, because detector architectures simultaneously optimize for classification, bounding box regression, and objectness estimation, their classification metrics are not directly comparable to those of LightTeaNet, which is optimized solely for image-level classification. For Parameter efficiency, the network integrates Depthwise Separable Convolutions, and for enhanced feature discrimination, it integrates Channel Attention. LightTeaNet has achieved a Precision of 0.9615, a Recall of 0.8772, and an F1-score of 0.9179, while it shows mAP@0.50=0.1810 without any manual annotations, reflecting the inherent coarseness of CAM-based localization under weak supervision. With only 2.32 M parameters, 6.11 GFLOPs, and a GPU inference latency of 2.44 ms (410.40 FPS), these results validate the model as an interpretable and computationally lightweight framework for intelligent disease monitoring in agriculture.