CPD-KD: a cooperative perception network for discrepancy feature fusion through knowledge distillation

Abstract The environmental perception of intelligent connected vehicles is a new perception paradigm centered around roadside sensing devices and 5G wireless communication. This approach expands the perception range of onboard sensors from localized observation to a global view, significantly enhanc...

Full description

Saved in:
Bibliographic Details
Main Authors: Caizhen He, Hai Wang, Tong Luo, Shunyao Zhang, Long Chen, Yingfeng Cai
Format: Article
Language:English
Published: Nature Portfolio 2025-07-01
Series:Scientific Reports
Subjects:
Online Access:https://doi.org/10.1038/s41598-025-08482-5
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849388238987329536
author Caizhen He
Hai Wang
Tong Luo
Shunyao Zhang
Long Chen
Yingfeng Cai
author_facet Caizhen He
Hai Wang
Tong Luo
Shunyao Zhang
Long Chen
Yingfeng Cai
author_sort Caizhen He
collection DOAJ
description Abstract The environmental perception of intelligent connected vehicles is a new perception paradigm centered around roadside sensing devices and 5G wireless communication. This approach expands the perception range of onboard sensors from localized observation to a global view, significantly enhancing the accuracy of vehicle perception. Traditional perception algorithms often face issues in real-world datasets, such as blurriness of target features from the vehicle’s perspective and loss of specific information when integrating vehicle and infrastructure features. To address these issues, we propose a Cooperative Perception network for Discrepancy feature fusion through Knowledge Distillation, named CPD-KD. The contributions of this work are threefold: First, we introduce a Sparse convolution-based Knowledge Distillation network (SKD) to alleviate blurring of single-view point cloud features by incorporating prior knowledge from fused viewpoint point cloud features. Second, we design a Discrepancy Feature Attention Fusion module to enhance Cooperative efficiency of differential information between vehicles and infrastructures. Finally, we validate the CPD-KD algorithm on the real-world dataset DAIR-V2X and the simulated dataset V2X-Set. Experimental results show that the proposed CPD-KD algorithm effectively enhances the accuracy of vehicular and infrastructural Cooperative perception.
format Article
id doaj-art-a63ba863cff7431d9e04ed5fa88ee145
institution Kabale University
issn 2045-2322
language English
publishDate 2025-07-01
publisher Nature Portfolio
record_format Article
series Scientific Reports
spelling doaj-art-a63ba863cff7431d9e04ed5fa88ee1452025-08-20T03:42:22ZengNature PortfolioScientific Reports2045-23222025-07-0115111410.1038/s41598-025-08482-5CPD-KD: a cooperative perception network for discrepancy feature fusion through knowledge distillationCaizhen He0Hai Wang1Tong Luo2Shunyao Zhang3Long Chen4Yingfeng Cai5School of Automotive and Traffic Engineering, Jiangsu UniversitySchool of Automotive and Traffic Engineering, Jiangsu UniversitySchool of Automobile and Traffic Engineering, Jiangsu University of TechnologyAutomotive Engineering Research Institute, Jiangsu UniversityAutomotive Engineering Research Institute, Jiangsu UniversityAutomotive Engineering Research Institute, Jiangsu UniversityAbstract The environmental perception of intelligent connected vehicles is a new perception paradigm centered around roadside sensing devices and 5G wireless communication. This approach expands the perception range of onboard sensors from localized observation to a global view, significantly enhancing the accuracy of vehicle perception. Traditional perception algorithms often face issues in real-world datasets, such as blurriness of target features from the vehicle’s perspective and loss of specific information when integrating vehicle and infrastructure features. To address these issues, we propose a Cooperative Perception network for Discrepancy feature fusion through Knowledge Distillation, named CPD-KD. The contributions of this work are threefold: First, we introduce a Sparse convolution-based Knowledge Distillation network (SKD) to alleviate blurring of single-view point cloud features by incorporating prior knowledge from fused viewpoint point cloud features. Second, we design a Discrepancy Feature Attention Fusion module to enhance Cooperative efficiency of differential information between vehicles and infrastructures. Finally, we validate the CPD-KD algorithm on the real-world dataset DAIR-V2X and the simulated dataset V2X-Set. Experimental results show that the proposed CPD-KD algorithm effectively enhances the accuracy of vehicular and infrastructural Cooperative perception.https://doi.org/10.1038/s41598-025-08482-5LiDAR sensorsKnowledge distillationV2X perception
spellingShingle Caizhen He
Hai Wang
Tong Luo
Shunyao Zhang
Long Chen
Yingfeng Cai
CPD-KD: a cooperative perception network for discrepancy feature fusion through knowledge distillation
Scientific Reports
LiDAR sensors
Knowledge distillation
V2X perception
title CPD-KD: a cooperative perception network for discrepancy feature fusion through knowledge distillation
title_full CPD-KD: a cooperative perception network for discrepancy feature fusion through knowledge distillation
title_fullStr CPD-KD: a cooperative perception network for discrepancy feature fusion through knowledge distillation
title_full_unstemmed CPD-KD: a cooperative perception network for discrepancy feature fusion through knowledge distillation
title_short CPD-KD: a cooperative perception network for discrepancy feature fusion through knowledge distillation
title_sort cpd kd a cooperative perception network for discrepancy feature fusion through knowledge distillation
topic LiDAR sensors
Knowledge distillation
V2X perception
url https://doi.org/10.1038/s41598-025-08482-5
work_keys_str_mv AT caizhenhe cpdkdacooperativeperceptionnetworkfordiscrepancyfeaturefusionthroughknowledgedistillation
AT haiwang cpdkdacooperativeperceptionnetworkfordiscrepancyfeaturefusionthroughknowledgedistillation
AT tongluo cpdkdacooperativeperceptionnetworkfordiscrepancyfeaturefusionthroughknowledgedistillation
AT shunyaozhang cpdkdacooperativeperceptionnetworkfordiscrepancyfeaturefusionthroughknowledgedistillation
AT longchen cpdkdacooperativeperceptionnetworkfordiscrepancyfeaturefusionthroughknowledgedistillation
AT yingfengcai cpdkdacooperativeperceptionnetworkfordiscrepancyfeaturefusionthroughknowledgedistillation