BT-YOLO11: Automatic Driving Road Target Detection in Complex Scenarios

Due to the problems faced by autonomous driving scenarios in complex scenarios, generalized target detectors are challenged, especially the detection accuracy of occluded and small targets. In order to solve this problem, a target detection algorithm for complex scenes BT-YOLO11 is proposed based on...

Full description

Saved in:
Bibliographic Details
Main Authors: Qi Wang, Qi Long Wang
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/10971358/
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849725082638745600
author Qi Wang
Qi Long Wang
author_facet Qi Wang
Qi Long Wang
author_sort Qi Wang
collection DOAJ
description Due to the problems faced by autonomous driving scenarios in complex scenarios, generalized target detectors are challenged, especially the detection accuracy of occluded and small targets. In order to solve this problem, a target detection algorithm for complex scenes BT-YOLO11 is proposed based on the latest YOLO11. In the feature extraction stage, Bi-level Routing Attention is introduced to enhance the model’s ability to capture feature information. The Tri-directional Feature Pyramid Net is used in the feature fusion stage, which adequately fuses different levels of feature information and improves the accuracy and robustness of the algorithm. By introducing an Adaptive Threshold Focus Loss function, the model focuses more on detection targets that are difficult to classify, the model generalization ability is improved. The experimental results show that the improved algorithm exhibits very competitive performance on the KITTI dataset. Specifically, it achieves a mAP50 metric of 95% and 77% under the more challenging mAP50:95 evaluation criterion. Compared to the YOLOv11s model, the algorithm proposed in this study improves the mAP50 by 2.6% and the mAP50:95 by 4.4%.
format Article
id doaj-art-c933ca8fbc684bfc8e23bef87b3fbfc3
institution DOAJ
issn 2169-3536
language English
publishDate 2025-01-01
publisher IEEE
record_format Article
series IEEE Access
spelling doaj-art-c933ca8fbc684bfc8e23bef87b3fbfc32025-08-20T03:10:34ZengIEEEIEEE Access2169-35362025-01-0113723647237410.1109/ACCESS.2025.356274710971358BT-YOLO11: Automatic Driving Road Target Detection in Complex ScenariosQi Wang0https://orcid.org/0000-0002-1404-582XQi Long Wang1https://orcid.org/0009-0001-6240-8058College of Computer Science, Nanjing University of Information Science & Technology, Nanjing, ChinaCollege of Computer Science, Nanjing University of Information Science & Technology, Nanjing, ChinaDue to the problems faced by autonomous driving scenarios in complex scenarios, generalized target detectors are challenged, especially the detection accuracy of occluded and small targets. In order to solve this problem, a target detection algorithm for complex scenes BT-YOLO11 is proposed based on the latest YOLO11. In the feature extraction stage, Bi-level Routing Attention is introduced to enhance the model’s ability to capture feature information. The Tri-directional Feature Pyramid Net is used in the feature fusion stage, which adequately fuses different levels of feature information and improves the accuracy and robustness of the algorithm. By introducing an Adaptive Threshold Focus Loss function, the model focuses more on detection targets that are difficult to classify, the model generalization ability is improved. The experimental results show that the improved algorithm exhibits very competitive performance on the KITTI dataset. Specifically, it achieves a mAP50 metric of 95% and 77% under the more challenging mAP50:95 evaluation criterion. Compared to the YOLOv11s model, the algorithm proposed in this study improves the mAP50 by 2.6% and the mAP50:95 by 4.4%.https://ieeexplore.ieee.org/document/10971358/Complex scenariosbi-level routing attentionKITTItri-directional feature pyramid netadaptive threshold focus lossYOLO11
spellingShingle Qi Wang
Qi Long Wang
BT-YOLO11: Automatic Driving Road Target Detection in Complex Scenarios
IEEE Access
Complex scenarios
bi-level routing attention
KITTI
tri-directional feature pyramid net
adaptive threshold focus loss
YOLO11
title BT-YOLO11: Automatic Driving Road Target Detection in Complex Scenarios
title_full BT-YOLO11: Automatic Driving Road Target Detection in Complex Scenarios
title_fullStr BT-YOLO11: Automatic Driving Road Target Detection in Complex Scenarios
title_full_unstemmed BT-YOLO11: Automatic Driving Road Target Detection in Complex Scenarios
title_short BT-YOLO11: Automatic Driving Road Target Detection in Complex Scenarios
title_sort bt yolo11 automatic driving road target detection in complex scenarios
topic Complex scenarios
bi-level routing attention
KITTI
tri-directional feature pyramid net
adaptive threshold focus loss
YOLO11
url https://ieeexplore.ieee.org/document/10971358/
work_keys_str_mv AT qiwang btyolo11automaticdrivingroadtargetdetectionincomplexscenarios
AT qilongwang btyolo11automaticdrivingroadtargetdetectionincomplexscenarios