CTB-YOLO: A lightweight framework for robust cotton terminal bud detection in complex field environments

Cotton terminal buds are critical for regulating plant architecture and determining yield formation, therefore, their automated detection is essential for intelligent cotton field cultivation. However, the ultra-small size of cotton buds, combined with complex illumination conditions and cluttered field backgrounds, poses significant challenges to robust real-time detection. To address these challenges, we present CTB-YOLO, a lightweight detection framework that integrates three complementary, purpose-built modules: the MSCA-FPN module (to prevent small-target feature dilution), the wConv2d module (to enhance illumination robustness), and the C2PSA-EDFFN module (to resolve contextual deficiency under severe occlusion). This integrated design ensures the framework systematically enhances fine-grained feature representation and multi-scale feature fusion for both efficiency and robustness, while significantly reducing model complexity and computational cost. On the self-constructed cotton terminal bud dataset, CTB-YOLO achieves a mean Average Precision (mAP50) of 85.3%, outperforming the YOLOv11n baseline by 2.3%, with 28.3% fewer parameters and 34.9% lower GFLOPs. Remarkably, CTB-YOLO delivers comparable accuracy to the transformer-based RT-DETR-R50 while requiring only 3.2% of its computational cost. Extensive evaluations further demonstrate strong robustness under strong sunlight, heavy occlusion, and low-contrast conditions, as well as competitive performance on the public Global Wheat Detection dataset under in-domain training. These results establish CTB-YOLO as a practical, efficient, and easily deployable framework for real-time small-target detection in intelligent cotton management and other precision agriculture applications.