Small Object Detection in Traffic Scenes Based on YOLO-MXANet

In terms of small objects in traffic scenes, general object detection algorithms have low detection accuracy, high model complexity, and slow detection speed. To solve the above problems, an improved algorithm (named YOLO-MXANet) is proposed in this paper. Complete-Intersection over Union (CIoU) is...

Descripción completa

Guardado en:
Detalles Bibliográficos
Autores principales: Xiaowei He, Rao Cheng, Zhonglong Zheng, Zeji Wang
Formato: article
Lenguaje:EN
Publicado: MDPI AG 2021
Materias:
Acceso en línea:https://doaj.org/article/b942a46497854050b3258785541471a8
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
id oai:doaj.org-article:b942a46497854050b3258785541471a8
record_format dspace
spelling oai:doaj.org-article:b942a46497854050b3258785541471a82021-11-11T19:20:13ZSmall Object Detection in Traffic Scenes Based on YOLO-MXANet10.3390/s212174221424-8220https://doaj.org/article/b942a46497854050b3258785541471a82021-11-01T00:00:00Zhttps://www.mdpi.com/1424-8220/21/21/7422https://doaj.org/toc/1424-8220In terms of small objects in traffic scenes, general object detection algorithms have low detection accuracy, high model complexity, and slow detection speed. To solve the above problems, an improved algorithm (named YOLO-MXANet) is proposed in this paper. Complete-Intersection over Union (CIoU) is utilized to improve loss function for promoting the positioning accuracy of the small object. In order to reduce the complexity of the model, we present a lightweight yet powerful backbone network (named SA-MobileNeXt) that incorporates channel and spatial attention. Our approach can extract expressive features more effectively by applying the Shuffle Channel and Spatial Attention (SCSA) module into the SandGlass Block (SGBlock) module while increasing the parameters by a small number. In addition, the data enhancement method combining Mosaic and Mixup is employed to improve the robustness of the training model. The Multi-scale Feature Enhancement Fusion (MFEF) network is proposed to fuse the extracted features better. In addition, the SiLU activation function is utilized to optimize the Convolution-Batchnorm-Leaky ReLU (CBL) module and the SGBlock module to accelerate the convergence of the model. The ablation experiments on the KITTI dataset show that each improved method is effective. The improved algorithm reduces the complexity and detection speed of the model while improving the object detection accuracy. The comparative experiments on the KITTY dataset and CCTSDB dataset with other algorithms show that our algorithm also has certain advantages.Xiaowei HeRao ChengZhonglong ZhengZeji WangMDPI AGarticledeep learningcomputer visionintelligence transportationYOLOv3lightweightChemical technologyTP1-1185ENSensors, Vol 21, Iss 7422, p 7422 (2021)
institution DOAJ
collection DOAJ
language EN
topic deep learning
computer vision
intelligence transportation
YOLOv3
lightweight
Chemical technology
TP1-1185
spellingShingle deep learning
computer vision
intelligence transportation
YOLOv3
lightweight
Chemical technology
TP1-1185
Xiaowei He
Rao Cheng
Zhonglong Zheng
Zeji Wang
Small Object Detection in Traffic Scenes Based on YOLO-MXANet
description In terms of small objects in traffic scenes, general object detection algorithms have low detection accuracy, high model complexity, and slow detection speed. To solve the above problems, an improved algorithm (named YOLO-MXANet) is proposed in this paper. Complete-Intersection over Union (CIoU) is utilized to improve loss function for promoting the positioning accuracy of the small object. In order to reduce the complexity of the model, we present a lightweight yet powerful backbone network (named SA-MobileNeXt) that incorporates channel and spatial attention. Our approach can extract expressive features more effectively by applying the Shuffle Channel and Spatial Attention (SCSA) module into the SandGlass Block (SGBlock) module while increasing the parameters by a small number. In addition, the data enhancement method combining Mosaic and Mixup is employed to improve the robustness of the training model. The Multi-scale Feature Enhancement Fusion (MFEF) network is proposed to fuse the extracted features better. In addition, the SiLU activation function is utilized to optimize the Convolution-Batchnorm-Leaky ReLU (CBL) module and the SGBlock module to accelerate the convergence of the model. The ablation experiments on the KITTI dataset show that each improved method is effective. The improved algorithm reduces the complexity and detection speed of the model while improving the object detection accuracy. The comparative experiments on the KITTY dataset and CCTSDB dataset with other algorithms show that our algorithm also has certain advantages.
format article
author Xiaowei He
Rao Cheng
Zhonglong Zheng
Zeji Wang
author_facet Xiaowei He
Rao Cheng
Zhonglong Zheng
Zeji Wang
author_sort Xiaowei He
title Small Object Detection in Traffic Scenes Based on YOLO-MXANet
title_short Small Object Detection in Traffic Scenes Based on YOLO-MXANet
title_full Small Object Detection in Traffic Scenes Based on YOLO-MXANet
title_fullStr Small Object Detection in Traffic Scenes Based on YOLO-MXANet
title_full_unstemmed Small Object Detection in Traffic Scenes Based on YOLO-MXANet
title_sort small object detection in traffic scenes based on yolo-mxanet
publisher MDPI AG
publishDate 2021
url https://doaj.org/article/b942a46497854050b3258785541471a8
work_keys_str_mv AT xiaoweihe smallobjectdetectionintrafficscenesbasedonyolomxanet
AT raocheng smallobjectdetectionintrafficscenesbasedonyolomxanet
AT zhonglongzheng smallobjectdetectionintrafficscenesbasedonyolomxanet
AT zejiwang smallobjectdetectionintrafficscenesbasedonyolomxanet
_version_ 1718431547440234496