为提高林业害虫识别精度,提出改进YOLOv7的林业害虫检测模型(GhostConv and SE attention enhanced YOLOv7,GS-YOLOv7)。首先,该模型将主干网络的传统卷积改用GhostConv轻量卷积,减小模型运行的参数量,提高模型效率;其次,通过添加挤压激...为提高林业害虫识别精度,提出改进YOLOv7的林业害虫检测模型(GhostConv and SE attention enhanced YOLOv7,GS-YOLOv7)。首先,该模型将主干网络的传统卷积改用GhostConv轻量卷积,减小模型运行的参数量,提高模型效率;其次,通过添加挤压激励(squeeze excitation,SE)注意力模块,强化对特征不显著的害虫图像边缘的提取能力,进而提高网络的特征提取能力;再次,用内容感知的特征重组(content aware reassembly of features,CARAFE)轻量级算子取代传统采样方法,提高特征重建质量,解决尺度不匹配问题,增强检测性能;最后,在Neck网络引入协调坐标卷积(coordinate convolution,CoordConv)模块,利用其位置信息解决目标定位不准问题,提高模型对空间位置的感受能力和泛化能力。在6种常见的病虫数据集上进行试验,GS-YOLOv7模型的精确率达到93.15%,交并比阈值为0.5时的平均精度均值达到93.29%,比原模型的精确率、平均精度均值分别提高6.50%和2.09%;参数量和模型大小分别降至1.9×10^(7)个和38.17 MB,比原模型分别降低51.4%和46.53%。结果表明,GS-YOLOv7模型较原模型性能有显著提升,模型改进有效。展开更多
随着智能交通系统的快速发展,交通标志检测在自动驾驶和辅助驾驶系统中扮演着至关重要的角色。为了应对交通标志检测对低延时和高精确度的要求,本文在YOLOv8n模型的基础上进行了改进,旨在提升检测速度和准确率。首先,从提升检测速度的...随着智能交通系统的快速发展,交通标志检测在自动驾驶和辅助驾驶系统中扮演着至关重要的角色。为了应对交通标志检测对低延时和高精确度的要求,本文在YOLOv8n模型的基础上进行了改进,旨在提升检测速度和准确率。首先,从提升检测速度的角度出发,本文提出了一种轻量化的卷积模块GhostConv,用于替代YOLOv8模型中的原始Conv模块,同时引入GhostC2f结构替换原始的C2f结构,以进一步减少计算复杂度并加速推理过程。其次,从提升检测准确率的角度出发,本文提出设计了SPPF-LSKA模块,增强了模型对交通标志特征的提取能力。此外,本文还提出使用SIou损失函数替换原始的Ciou损失函数,以更好地优化边界框回归,进一步提升检测精度。实验结果表明,改进后的模型在保持较低延时的同时,显著提升了交通标志检测的准确率,能够更好地满足实际应用需求。With the rapid development of intelligent transportation systems, traffic sign detection plays a critical role in autonomous and assisted driving systems. To address the requirements of low latency and high accuracy in traffic sign detection, this paper improves the YOLOv8n model by focusing on enhancing detection speed and accuracy. First, from the perspective of improving detection speed, we propose a lightweight GhostConv module to replace the original Conv module in YOLOv8, and introduce the GhostC2f structure to substitute the original C2f structure, thereby reducing computational complexity and accelerating inference. Second, in this paper, the sppf-lska module is designed to enhance the ability of the model to extract the characteristics of traffic signs. Additionally, we employ the SIoU loss function instead of the original CIoU loss function to optimize bounding box regression and further improve detection precision. Experimental results demonstrate that the improved model achieves significantly higher accuracy while maintaining low latency, better meeting the demands of real-world applications.展开更多
文摘为提高林业害虫识别精度,提出改进YOLOv7的林业害虫检测模型(GhostConv and SE attention enhanced YOLOv7,GS-YOLOv7)。首先,该模型将主干网络的传统卷积改用GhostConv轻量卷积,减小模型运行的参数量,提高模型效率;其次,通过添加挤压激励(squeeze excitation,SE)注意力模块,强化对特征不显著的害虫图像边缘的提取能力,进而提高网络的特征提取能力;再次,用内容感知的特征重组(content aware reassembly of features,CARAFE)轻量级算子取代传统采样方法,提高特征重建质量,解决尺度不匹配问题,增强检测性能;最后,在Neck网络引入协调坐标卷积(coordinate convolution,CoordConv)模块,利用其位置信息解决目标定位不准问题,提高模型对空间位置的感受能力和泛化能力。在6种常见的病虫数据集上进行试验,GS-YOLOv7模型的精确率达到93.15%,交并比阈值为0.5时的平均精度均值达到93.29%,比原模型的精确率、平均精度均值分别提高6.50%和2.09%;参数量和模型大小分别降至1.9×10^(7)个和38.17 MB,比原模型分别降低51.4%和46.53%。结果表明,GS-YOLOv7模型较原模型性能有显著提升,模型改进有效。
文摘随着智能交通系统的快速发展,交通标志检测在自动驾驶和辅助驾驶系统中扮演着至关重要的角色。为了应对交通标志检测对低延时和高精确度的要求,本文在YOLOv8n模型的基础上进行了改进,旨在提升检测速度和准确率。首先,从提升检测速度的角度出发,本文提出了一种轻量化的卷积模块GhostConv,用于替代YOLOv8模型中的原始Conv模块,同时引入GhostC2f结构替换原始的C2f结构,以进一步减少计算复杂度并加速推理过程。其次,从提升检测准确率的角度出发,本文提出设计了SPPF-LSKA模块,增强了模型对交通标志特征的提取能力。此外,本文还提出使用SIou损失函数替换原始的Ciou损失函数,以更好地优化边界框回归,进一步提升检测精度。实验结果表明,改进后的模型在保持较低延时的同时,显著提升了交通标志检测的准确率,能够更好地满足实际应用需求。With the rapid development of intelligent transportation systems, traffic sign detection plays a critical role in autonomous and assisted driving systems. To address the requirements of low latency and high accuracy in traffic sign detection, this paper improves the YOLOv8n model by focusing on enhancing detection speed and accuracy. First, from the perspective of improving detection speed, we propose a lightweight GhostConv module to replace the original Conv module in YOLOv8, and introduce the GhostC2f structure to substitute the original C2f structure, thereby reducing computational complexity and accelerating inference. Second, in this paper, the sppf-lska module is designed to enhance the ability of the model to extract the characteristics of traffic signs. Additionally, we employ the SIoU loss function instead of the original CIoU loss function to optimize bounding box regression and further improve detection precision. Experimental results demonstrate that the improved model achieves significantly higher accuracy while maintaining low latency, better meeting the demands of real-world applications.
