C2PSA-Enhanced YOLOv11 Architecture: A Novel Approach for Small Target Detection in Cotton Disease Diagnosis

2025-08-16

AI Summary

This paper addresses three critical challenges in cotton disease detection: high miss-rate (35%) for early-stage small targets (<5 mm²), poor robustness under field conditions (25% accuracy drop), and high multi-disease confusion error (34.7%). To tackle these, we propose a lightweight YOLOv11-based detection framework. Our method introduces the C2PSA module to enhance small-target feature representation, employs a dynamic class-weighted loss to mitigate class imbalance, and refines the Mosaic-MixUp data augmentation strategy to improve generalization on low-quality images and across multiple diseases. Evaluated on 4,078 real-world field images, our approach achieves an mAP₅₀ of 0.820 (+8.0% over baseline) and mAP₅₀₋₉₅ of 0.705 (+10.5%), with real-time inference at 158 FPS—enabling efficient deployment on mobile devices. The framework significantly advances practical, field-deployable intelligent monitoring for precision agriculture.

Technology Category

• Computer Vision: Object Detection & Categorization
• Machine Learning: Hardware-aware ML
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

This study presents a deep learning-based optimization of YOLOv11 for cotton disease detection, developing an intelligent monitoring system. Three key challenges are addressed: (1) low precision in early spot detection (35% leakage rate for sub-5mm2 spots), (2) performance degradation in field conditions (25% accuracy drop), and (3) high error rates (34.7%) in multi-disease scenarios. The proposed solutions include: C2PSA module for enhanced small-target feature extraction; Dynamic category weighting to handle sample imbalance; Improved data augmentation via Mosaic-MixUp scaling. Experimental results on a 4,078-image dataset show: mAP50: 0.820 (+8.0% improvement); mAP50-95: 0.705 (+10.5% improvement); Inference speed: 158 FPS. The mobile-deployed system enables real-time disease monitoring and precision treatment in agricultural applications.

Problem

• Low precision in early small cotton disease spot detection
• Performance degradation under field conditions
• High error rates in multi-disease classification scenarios

Innovation

• C2PSA module enhances small-target feature extraction
• Dynamic category weighting handles sample imbalance
• Improved data augmentation via Mosaic-MixUp scaling