Related Experiment Video
Updated: Aug 5, 2026
05:28
Visualization of Leaf and Bracteal Nectaries of Cotton using Digital Microscopy to Improve Scoring Accuracy and Data Preservation
Published on: February 6, 2026
Cotton Leaf Spot Detection Based on an Improved YOLOv11n Model
Yaxin Xie1, Mingyu Zhang1, Yonghua Han1
1School of Information Science and Engineering, Zhejiang Sci-Tech University, Hangzhou 310018, China.
Journal of Imaging
|July 27, 2026
Summary
This study introduces an improved YOLOv11n algorithm for accurate cotton disease detection, especially for small lesions. The enhanced model boosts detection accuracy while maintaining efficient inference for real-time agricultural monitoring.
Area of Science:
- Agricultural Science
- Computer Vision
- Machine Learning
Background:
- Detecting cotton diseases is challenging due to complex environments and varying lesion scales, particularly small targets.
- Lightweight deep learning models often struggle with accuracy in these conditions.
Purpose of the Study:
- To enhance the accuracy and efficiency of lightweight models for cotton disease detection.
- To address the limitations of current models in identifying small and low-resolution disease spots.
Main Methods:
- Reconstruction of the backbone network using the GhostConv (G-conv) module to reduce computational complexity.
- Development of an Adaptive Calibration and Feature Fusion Architecture Head (ACFFA) for handling scale variations.
- Introduction of an Adaptive Scale-aware Wise Intersection over Union (AS-WIoU) loss function to improve bounding box regression and small-target detection.
Main Results:
- The proposed YOLOv11n model achieved mAP50 of 90.30% and mAP50-95 of 73.84% on a cotton disease dataset.
- Precision reached 92.33% and recall reached 87.68%, with a parameter count of 3.81 M.
- Significant improvements in detection accuracy were observed, especially for small and low-resolution lesions.
Conclusions:
- The improved YOLOv11n algorithm effectively enhances cotton disease detection accuracy.
- The model's efficiency makes it suitable for real-time monitoring on agricultural embedded systems.
