YOLO-AFR: An Improved YOLOv12-Based Model for Accurate and Real-Time Dangerous Driving Behavior Detection

Accurate detection of dangerous driving behaviors is crucial for improving the safety of intelligent transportation systems. However, existing methods often struggle with limited feature extraction capabilities and insufficient attention to multiscale and contextual information. To overcome these li...

Full description

Saved in:
Bibliographic Details
Main Authors: Tianchen Ge, Bo Ning, Yiwu Xie
Format: Article
Language:English
Published: MDPI AG 2025-05-01
Series:Applied Sciences
Subjects:
Online Access:https://www.mdpi.com/2076-3417/15/11/6090
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850129354846109696
author Tianchen Ge
Bo Ning
Yiwu Xie
author_facet Tianchen Ge
Bo Ning
Yiwu Xie
author_sort Tianchen Ge
collection DOAJ
description Accurate detection of dangerous driving behaviors is crucial for improving the safety of intelligent transportation systems. However, existing methods often struggle with limited feature extraction capabilities and insufficient attention to multiscale and contextual information. To overcome these limitations, we propose YOLO-AFR (YOLO with Adaptive Feature Refinement) for dangerous driving behavior detection. YOLO-AFR builds upon the YOLOv12 architecture and introduces three key innovations: (1) the redesign of the original A2C2f module by introducing a Feature-Refinement Feedback Network (FRFN), resulting in a new A2C2f-FRFN structure that adaptively refines multiscale features, (2) the integration of self-calibrated convolution (SC-Conv) modules in the backbone to enhance multiscale contextual modeling, and (3) the employment of a SEAM-based detection head to improve global contextual awareness and prediction accuracy. These three modules combine to form a Calibration-Refinement Loop, which progressively reduces redundancy and enhances discriminative features layer by layer. We evaluate YOLO-AFR on two public driver behavior datasets, YawDD-E and SfdDD. Experimental results show that YOLO-AFR significantly outperforms the baseline YOLOv12 model, achieving improvements of 1.3% and 1.8% in mAP@0.5, and 2.6% and 12.3% in mAP@0.5:0.95 on the YawDD-E and SfdDD datasets, respectively, demonstrating its superior performance in complex driving scenarios while maintaining high inference speed.
format Article
id doaj-art-d2fe72417ea949df8af1e3b0ecabae9d
institution OA Journals
issn 2076-3417
language English
publishDate 2025-05-01
publisher MDPI AG
record_format Article
series Applied Sciences
spelling doaj-art-d2fe72417ea949df8af1e3b0ecabae9d2025-08-20T02:33:01ZengMDPI AGApplied Sciences2076-34172025-05-011511609010.3390/app15116090YOLO-AFR: An Improved YOLOv12-Based Model for Accurate and Real-Time Dangerous Driving Behavior DetectionTianchen Ge0Bo Ning1Yiwu Xie2Information Science and Technology College, Dalian Maritime University, Dalian 116026, ChinaInformation Science and Technology College, Dalian Maritime University, Dalian 116026, ChinaInformation Science and Technology College, Dalian Maritime University, Dalian 116026, ChinaAccurate detection of dangerous driving behaviors is crucial for improving the safety of intelligent transportation systems. However, existing methods often struggle with limited feature extraction capabilities and insufficient attention to multiscale and contextual information. To overcome these limitations, we propose YOLO-AFR (YOLO with Adaptive Feature Refinement) for dangerous driving behavior detection. YOLO-AFR builds upon the YOLOv12 architecture and introduces three key innovations: (1) the redesign of the original A2C2f module by introducing a Feature-Refinement Feedback Network (FRFN), resulting in a new A2C2f-FRFN structure that adaptively refines multiscale features, (2) the integration of self-calibrated convolution (SC-Conv) modules in the backbone to enhance multiscale contextual modeling, and (3) the employment of a SEAM-based detection head to improve global contextual awareness and prediction accuracy. These three modules combine to form a Calibration-Refinement Loop, which progressively reduces redundancy and enhances discriminative features layer by layer. We evaluate YOLO-AFR on two public driver behavior datasets, YawDD-E and SfdDD. Experimental results show that YOLO-AFR significantly outperforms the baseline YOLOv12 model, achieving improvements of 1.3% and 1.8% in mAP@0.5, and 2.6% and 12.3% in mAP@0.5:0.95 on the YawDD-E and SfdDD datasets, respectively, demonstrating its superior performance in complex driving scenarios while maintaining high inference speed.https://www.mdpi.com/2076-3417/15/11/6090dangerous driving behavior detectionYOLOfeature refinementself-calibrated convolutionseparated and enhancement attentiondeep learning
spellingShingle Tianchen Ge
Bo Ning
Yiwu Xie
YOLO-AFR: An Improved YOLOv12-Based Model for Accurate and Real-Time Dangerous Driving Behavior Detection
Applied Sciences
dangerous driving behavior detection
YOLO
feature refinement
self-calibrated convolution
separated and enhancement attention
deep learning
title YOLO-AFR: An Improved YOLOv12-Based Model for Accurate and Real-Time Dangerous Driving Behavior Detection
title_full YOLO-AFR: An Improved YOLOv12-Based Model for Accurate and Real-Time Dangerous Driving Behavior Detection
title_fullStr YOLO-AFR: An Improved YOLOv12-Based Model for Accurate and Real-Time Dangerous Driving Behavior Detection
title_full_unstemmed YOLO-AFR: An Improved YOLOv12-Based Model for Accurate and Real-Time Dangerous Driving Behavior Detection
title_short YOLO-AFR: An Improved YOLOv12-Based Model for Accurate and Real-Time Dangerous Driving Behavior Detection
title_sort yolo afr an improved yolov12 based model for accurate and real time dangerous driving behavior detection
topic dangerous driving behavior detection
YOLO
feature refinement
self-calibrated convolution
separated and enhancement attention
deep learning
url https://www.mdpi.com/2076-3417/15/11/6090
work_keys_str_mv AT tianchenge yoloafranimprovedyolov12basedmodelforaccurateandrealtimedangerousdrivingbehaviordetection
AT boning yoloafranimprovedyolov12basedmodelforaccurateandrealtimedangerousdrivingbehaviordetection
AT yiwuxie yoloafranimprovedyolov12basedmodelforaccurateandrealtimedangerousdrivingbehaviordetection