当前位置: X-MOL 学术IEEE Access › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
An Improved YOLOv5 Method for Small Object Detection in UAV Capture Scenes
IEEE Access ( IF 3.9 ) Pub Date : 2023-01-31 , DOI: 10.1109/access.2023.3241005
Zhen Liu 1 , Xuehui Gao 1 , Yu Wan 1 , Jianhao Wang 1 , Hao Lyu 1
Affiliation  

Aiming at the problem of a large number of small dense objects in high-altitude shooting and complex background noise interference in the captured scenes, an improved object detection algorithm for YOLOv5 UAV capture scenes is proposed. A Feature Enhancement Block (FEBlock) is first proposed to generate adaptive weights for different receptive field features by convolution, assigning major weights to shallow feature maps to improve small object feature extraction ability. The FEBlock is then integrated into Spatial Pyramid Pooling (SPP) to generate Enhanced Spatial Pyramid Pooling (ESPP), which performs feature enhancement for the result of each maximum pooling; and creates new features containing multi-scale contextual information with better feature characterization capability by weighting fused contextual features. Secondly, the Self-Characteristic Expansion Plate (SCEP) is proposed, which achieves the fusion and expansion of feature information through compression, non-linear mapping, and expansion with its own module, further improving the network’s capacity for feature extraction and generating a new spatial pyramid pooling (ESPP-S) by splicing with ESPP. Finally, a shallower feature map is added as a detection layer to the YOLOv5 network model’s large, medium, and small detection layers to improve the network’s detection performance for medium and long-range objects. Experiments were conducted on the VisDrone2021 dataset, and the results showed that the improved YOLOv5 model improved mAP0.5 by 4.6%, mAP0.5:0.95 by 2.9%, and precision by 2.7%. The mAP0.5 of the model trained at the input resolution of $1024\times1024$ reached 56.8%. The experiments show that the improved YOLOv5 model can improve object detection accuracy for UAV capture scenes.

中文翻译:

一种改进的 YOLOv5 无人机捕捉场景小目标检测方法

【摘要】:针对高空拍摄中存在大量小而致密物体、捕捉场景背景噪声干扰复杂等问题,提出一种改进的YOLOv5无人机捕捉场景物体检测算法。首先提出了一种特征增强块(FEBlock),通过卷积为不同的感受野特征生成自适应权重,将主要权重分配给浅层特征图,以提高小物体特征提取能力。然后将FEBlock集成到Spatial Pyramid Pooling (SPP)中生成Enhanced Spatial Pyramid Pooling (ESPP),对每个maximum pooling的结果进行特征增强;并通过加权融合的上下文特征来创建包含多尺度上下文信息的新特征,具有更好的特征表征能力。第二,提出了Self-Characteristic Expansion Plate(SCEP),通过自身模块的压缩、非线性映射和扩展实现特征信息的融合和扩展,进一步提高网络的特征提取能力,生成新的空间金字塔通过与 ESPP 拼接进行池化 (ESPP-S)。最后在YOLOv5网络模型的大、中、小检测层中加入一个较浅的特征图作为检测层,以提高网络对中远距离物体的检测性能。在VisDrone2021数据集上进行了实验,结果表明,改进后的YOLOv5模型mAP0.5提高了4.6%,mAP0.5:0.95提高了2.9%,精度提高了2.7%。在输入分辨率下训练的模型的 mAP0.5 它通过压缩、非线性映射和自身模块的扩展实现了特征信息的融合和扩展,进一步提高了网络的特征提取能力,并通过与ESPP的拼接生成了新的空间金字塔池化(ESPP-S)。最后在YOLOv5网络模型的大、中、小检测层中加入一个较浅的特征图作为检测层,以提高网络对中远距离物体的检测性能。在VisDrone2021数据集上进行了实验,结果表明,改进后的YOLOv5模型mAP0.5提高了4.6%,mAP0.5:0.95提高了2.9%,精度提高了2.7%。在输入分辨率下训练的模型的 mAP0.5 它通过压缩、非线性映射和自身模块的扩展实现了特征信息的融合和扩展,进一步提高了网络的特征提取能力,并通过与ESPP的拼接生成了新的空间金字塔池化(ESPP-S)。最后在YOLOv5网络模型的大、中、小检测层中加入一个较浅的特征图作为检测层,以提高网络对中远距离物体的检测性能。在VisDrone2021数据集上进行了实验,结果表明,改进后的YOLOv5模型mAP0.5提高了4.6%,mAP0.5:0.95提高了2.9%,精度提高了2.7%。在输入分辨率下训练的模型的 mAP0.5 进一步提高网络的特征提取能力,通过与ESPP拼接生成新的空间金字塔池化(ESPP-S)。最后在YOLOv5网络模型的大、中、小检测层中加入一个较浅的特征图作为检测层,以提高网络对中远距离物体的检测性能。在VisDrone2021数据集上进行了实验,结果表明,改进后的YOLOv5模型mAP0.5提高了4.6%,mAP0.5:0.95提高了2.9%,精度提高了2.7%。在输入分辨率下训练的模型的 mAP0.5 进一步提高网络的特征提取能力,通过与ESPP拼接生成新的空间金字塔池化(ESPP-S)。最后在YOLOv5网络模型的大、中、小检测层中加入一个较浅的特征图作为检测层,以提高网络对中远距离物体的检测性能。在VisDrone2021数据集上进行了实验,结果表明,改进后的YOLOv5模型mAP0.5提高了4.6%,mAP0.5:0.95提高了2.9%,精度提高了2.7%。在输入分辨率下训练的模型的 mAP0.5 和小检测层,以提高网络对中远程目标的检测性能。在VisDrone2021数据集上进行了实验,结果表明,改进后的YOLOv5模型mAP0.5提高了4.6%,mAP0.5:0.95提高了2.9%,精度提高了2.7%。在输入分辨率下训练的模型的 mAP0.5 和小检测层,以提高网络对中远程目标的检测性能。在VisDrone2021数据集上进行了实验,结果表明,改进后的YOLOv5模型mAP0.5提高了4.6%,mAP0.5:0.95提高了2.9%,精度提高了2.7%。在输入分辨率下训练的模型的 mAP0.5 $1024\times1024$达到56.8%。实验表明,改进后的YOLOv5模型可以提高无人机捕捉场景的目标检测精度。
更新日期:2023-01-31
down
wechat
bug