Abstract: To address challenges in underwater small object detection, such as light scattering, low contrast, and complex backgrounds, an improved algorithm named CSAF-YOLO (Cross-Scale Adaptive Fusion YOLO) was proposed based on YOLO11. Firstly, a Multi-Scale Collaborative Fusion (MSCF) module was designed to enhance cross-scale feature synergy and contextual information extraction through spatial fusion and channel interaction mechanisms. Secondly, a Dynamic Kernel Scale Modulation (DKSM) module was constructed to adaptively generate local and global modulation matrices, optimizing convolutional kernels for improved adaptability to complex underwater environments. Thirdly, a Multi-Scale Enhanced Detection Head (MSE-Head) was proposed to improve small object localization accuracy via scale-aware enhancement and dynamic cross-scale feature fusion. Finally, the Modified Penalized Distance Intersection over Union (MPDIoU) loss function was introduced to optimize bounding box regression for underwater small objects through minimum point distance and multi-scale penalty mechanisms. Experimental results on the URPC2020 dataset demonstrate that CSAF-YOLO achieves an mAP50 (mean Average Precision at 50% Intersection over Union (IoU) threshold) of 85.0%, representing a 1.6% improvement over YOLO11. The proposed algorithm provides an effective solution for visual tasks in fields such as marine resource exploration and underwater robotic navigation.

Key words: underwater small object detection, YOLO11, multi-scale feature fusion, dynamic kernel modulation, attention mechanism

摘要： 针对水下小目标检测中光线散射、低对比度和复杂背景等挑战，提出一种基于YOLO11的改进算法CSAF-YOLO(Cross-Scale Adaptive Fusion YOLO)。首先，设计多尺度协同融合模块(MSCF)，通过空间融合与通道交互机制，增强多尺度特征间的协同作用，提升上下文信息提取能力；其次，构建动态内核尺度调制模块(DKSM)，自适应生成局部与全局调制矩阵，优化卷积核以增强模型对复杂水下环境的适应性；再次，提出多尺度增强检测头(MSE-Head)，通过尺度感知增强和跨尺度特征动态融合，提高小目标定位精度；最后，引入MPDIoU(Modified Penalized Distance Intersection over Union)损失函数，通过最小点距离和多尺度惩罚机制，优化水下小型目标的边界框回归。在URPC2020数据集上的实验结果表明，CSAF-YOLO在50%交并比(IoU)阈值下的平均精度均值(mAP50)达到了85.0%，相较于YOLO11提升了1.6%，为海洋资源勘探和水下机器人导航等领域的视觉任务提供有效的解决方案。

关键词: 水下小目标检测, YOLO11, 多尺度特征融合, 动态内核调制, 注意力机制