首页|改进融合注意力机制的小目标和被遮挡目标检测

改进融合注意力机制的小目标和被遮挡目标检测

扫码查看
针对夜间光照条件不足等条件下交通环境的多目标检测问题,提出一种改进YOLOv5s的目标检测算法.该算法首先在原始的YOLOv5s网络中嵌入三分支结构并行卷积注意力模块,通过计算跨维度注意力权值矩阵,实现了一种轻量级的有效注意力机制.其次,为了解决小目标和遮挡目标的检测问题,嵌入残差遮挡感知注意力机制,通过不同卷积核大小的卷积块对图像进行类分块操作,更准确地突显小目标和被遮挡目标.通过在FLIR数据集上的对比实验表明,改进算法在夜间交通环境下的多目标检测任务中能够提高检测精度,相较于传统YOLOv5s,其检测准确率mAP@0.5提高2.9%.
Improved fusion attention mechanism for detecting small and occluded targets
A modified YOLOv5s object detection algorithm is proposed to address the problem of multi-target detection in traffic environments under conditions such as insufficient nighttime lighting.This algorithm first embeds a three branch parallel convolutional attention module into the original YOLOv5s network,and achieves a lightweight and effective attention mechanism by calculating the cross dimensional attention weight matrix.Secondly,in order to solve the detection problem of small and occluded targets,a residual occlusion perception attention mechanism is embedded.The image is segmented into different convolution blocks with different kernel sizes to more accurately highlight small and occluded targets.Comparative experiments on the FL1R dataset show that this improved algorithm can improve detection accuracy in multi-object detection tasks in nighttime traffic environments,and its detection accuracy is higher than that of traditional YOLOv5s map@.5 Increase by 2.9%.

target detectionYOLOv5triple parallel convolutional attention mechanismresidual occlusion attention

刘丽伟、王玲、戚星烁

展开 >

长春工业大学计算机科学与工程学院,吉林 长春 130102

目标检测 YOLOv5 三重并行卷积注意力机制 残差遮挡注意力机制

2024

长春工业大学学报
长春工业大学

长春工业大学学报

影响因子:0.282
ISSN:1674-1374
年,卷(期):2024.45(5)