ACNet: An Attention–Convolution Collaborative Semantic Segmentation Network on Sensor-Derived Datasets for Autonomous Driving

In intelligent vehicular networks, the accuracy of semantic segmentation in road scenes is crucial for vehicle-mounted artificial intelligence to achieve environmental perception, decision support, and safety control. Although deep learning methods have made significant progress, two main challenges...

Full description

Saved in:
Bibliographic Details
Main Authors: Qiliang Zhang, Kaiwen Hua, Zi Zhang, Yiwei Zhao, Pengpeng Chen
Format: Article
Language:English
Published: MDPI AG 2025-08-01
Series:Sensors
Subjects:
Online Access:https://www.mdpi.com/1424-8220/25/15/4776
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849239772740976640
author Qiliang Zhang
Kaiwen Hua
Zi Zhang
Yiwei Zhao
Pengpeng Chen
author_facet Qiliang Zhang
Kaiwen Hua
Zi Zhang
Yiwei Zhao
Pengpeng Chen
author_sort Qiliang Zhang
collection DOAJ
description In intelligent vehicular networks, the accuracy of semantic segmentation in road scenes is crucial for vehicle-mounted artificial intelligence to achieve environmental perception, decision support, and safety control. Although deep learning methods have made significant progress, two main challenges remain: first, the difficulty in balancing global and local features leads to blurred object boundaries and misclassification; second, conventional convolutions have limited ability to perceive irregular objects, causing information loss and affecting segmentation accuracy. To address these issues, this paper proposes a global–local collaborative attention module and a spider web convolution module. The former enhances feature representation through bidirectional feature interaction and dynamic weight allocation, reducing false positives and missed detections. The latter introduces an asymmetric sampling topology and six-directional receptive field paths to effectively improve the recognition of irregular objects. Experiments on the Cityscapes, CamVid, and BDD100K datasets, collected using vehicle-mounted cameras, demonstrate that the proposed method performs excellently across multiple evaluation metrics, including mIoU, mRecall, mPrecision, and mAccuracy. Comparative experiments with classical segmentation networks, attention mechanisms, and convolution modules validate the effectiveness of the proposed approach. The proposed method demonstrates outstanding performance in sensor-based semantic segmentation tasks and is well-suited for environmental perception systems in autonomous driving.
format Article
id doaj-art-e7cd27de082c4f85b9e05a435b02c6fa
institution Kabale University
issn 1424-8220
language English
publishDate 2025-08-01
publisher MDPI AG
record_format Article
series Sensors
spelling doaj-art-e7cd27de082c4f85b9e05a435b02c6fa2025-08-20T04:00:50ZengMDPI AGSensors1424-82202025-08-012515477610.3390/s25154776ACNet: An Attention–Convolution Collaborative Semantic Segmentation Network on Sensor-Derived Datasets for Autonomous DrivingQiliang Zhang0Kaiwen Hua1Zi Zhang2Yiwei Zhao3Pengpeng Chen4School of Computer Science and Technology, China University of Mining and Technology, Xuzhou 221116, ChinaSchool of Computer Science and Technology, China University of Mining and Technology, Xuzhou 221116, ChinaSchool of Computer Science and Technology, China University of Mining and Technology, Xuzhou 221116, ChinaSchool of Mechanical, Electronic and Control Engineering, Beijing Jiaotong University, Beijing 100044, ChinaSchool of Computer Science and Technology, China University of Mining and Technology, Xuzhou 221116, ChinaIn intelligent vehicular networks, the accuracy of semantic segmentation in road scenes is crucial for vehicle-mounted artificial intelligence to achieve environmental perception, decision support, and safety control. Although deep learning methods have made significant progress, two main challenges remain: first, the difficulty in balancing global and local features leads to blurred object boundaries and misclassification; second, conventional convolutions have limited ability to perceive irregular objects, causing information loss and affecting segmentation accuracy. To address these issues, this paper proposes a global–local collaborative attention module and a spider web convolution module. The former enhances feature representation through bidirectional feature interaction and dynamic weight allocation, reducing false positives and missed detections. The latter introduces an asymmetric sampling topology and six-directional receptive field paths to effectively improve the recognition of irregular objects. Experiments on the Cityscapes, CamVid, and BDD100K datasets, collected using vehicle-mounted cameras, demonstrate that the proposed method performs excellently across multiple evaluation metrics, including mIoU, mRecall, mPrecision, and mAccuracy. Comparative experiments with classical segmentation networks, attention mechanisms, and convolution modules validate the effectiveness of the proposed approach. The proposed method demonstrates outstanding performance in sensor-based semantic segmentation tasks and is well-suited for environmental perception systems in autonomous driving.https://www.mdpi.com/1424-8220/25/15/4776semantic segmentationdeep learningattention mechanismconvolutionvehicle-mounted camerasautonomous driving
spellingShingle Qiliang Zhang
Kaiwen Hua
Zi Zhang
Yiwei Zhao
Pengpeng Chen
ACNet: An Attention–Convolution Collaborative Semantic Segmentation Network on Sensor-Derived Datasets for Autonomous Driving
Sensors
semantic segmentation
deep learning
attention mechanism
convolution
vehicle-mounted cameras
autonomous driving
title ACNet: An Attention–Convolution Collaborative Semantic Segmentation Network on Sensor-Derived Datasets for Autonomous Driving
title_full ACNet: An Attention–Convolution Collaborative Semantic Segmentation Network on Sensor-Derived Datasets for Autonomous Driving
title_fullStr ACNet: An Attention–Convolution Collaborative Semantic Segmentation Network on Sensor-Derived Datasets for Autonomous Driving
title_full_unstemmed ACNet: An Attention–Convolution Collaborative Semantic Segmentation Network on Sensor-Derived Datasets for Autonomous Driving
title_short ACNet: An Attention–Convolution Collaborative Semantic Segmentation Network on Sensor-Derived Datasets for Autonomous Driving
title_sort acnet an attention convolution collaborative semantic segmentation network on sensor derived datasets for autonomous driving
topic semantic segmentation
deep learning
attention mechanism
convolution
vehicle-mounted cameras
autonomous driving
url https://www.mdpi.com/1424-8220/25/15/4776
work_keys_str_mv AT qiliangzhang acnetanattentionconvolutioncollaborativesemanticsegmentationnetworkonsensorderiveddatasetsforautonomousdriving
AT kaiwenhua acnetanattentionconvolutioncollaborativesemanticsegmentationnetworkonsensorderiveddatasetsforautonomousdriving
AT zizhang acnetanattentionconvolutioncollaborativesemanticsegmentationnetworkonsensorderiveddatasetsforautonomousdriving
AT yiweizhao acnetanattentionconvolutioncollaborativesemanticsegmentationnetworkonsensorderiveddatasetsforautonomousdriving
AT pengpengchen acnetanattentionconvolutioncollaborativesemanticsegmentationnetworkonsensorderiveddatasetsforautonomousdriving