CID-YOLO: enhancing YOLO for cotton impurity detection
Yi Wang, Jupu Wang, Menghui Li, Zhenduo Guo, Yan Dong
Zhongyuan University of Technology University of Electronic Science and Technology of China
阅读操作
确认中在文库中上传 PDF 后可生成中文音频讲解。
摘要与影响
Automated cotton impurity detection is pivotal for quality grading in textile processing. However, in real-world industrial scenarios, impurities often exhibit irregular morphologies and low contrast against densely textured cotton fibers, which results in insufficient feature representation of generic object detectors, and makes them ineffective in such cluttered environments. To address this problem, we propose an enhanced YOLO for cotton impurity detection, namely CID-YOLO. Built upon the YOLO framework, our method enhances discriminative capability through three strategic designs: Hybrid Global-Local Collaborative Modeling to capture multi-scale contextual cues, Safe Cross-Stage Attention C2f to refine feature interactions, and Scale-Focused Detail Injection Network to optimize detection across varying target sizes. Extensive experiments on a self-built Cotton Impurity Detection dataset demonstrate that our method achieves 30.3% mAP0.5 and 49.5% precision, significantly outperforming the YOLOv8s baseline. With negligible computational overhead, the proposed model effectively suppresses false positives in fiber-rich scenes, offering a practical solution for high-precision industrial inspection.
逐年被引趋势
暂无年度引用数据
关键指标
同类平均 = 1
同领域 · 同年份 · 同类型
Google Scholar 与 OpenAlex 的被引统计范围不同,数值存在差异属正常。
AI 辅助阅读
依据:摘要
可就本文提问;依据不足时会说明。
学术脉络
学科主题
工程Industrial Vision Systems and Defect Detection
Advanced Neural Network Applications · Smart Agriculture and AI