Traffic Sign Recognition for Autonomous Driving Using Branched YOLOv2 and Geometric Features

Arefeh Rezaei

cs.CVSep 18, 2026

Traffic Sign Recognition for Autonomous Driving Using Branched YOLOv2 and Geometric Features

Authors: Arefeh Rezaei

Organizations: Department of Computer Engineering, K. N. Toosi University of Technology

Abstract

Traffic sign recognition (TSR) is an important perception task for autonomous driving and advanced driver-assistance systems, where a system must both localize traffic signs and determine their semantic classes efficiently. This work presents a TSR system based on YOLOv2 for simultaneous detection and classification. Two complementary modifications are studied. First, YOLOv2 is extended with intermediate prediction layers, forming a branched architecture that can terminate inference early for easy cases and reduce computation time. Both whole-image and cell-wise branching strategies are investigated. Second, geometric information is introduced to reduce classification errors between visually similar signs. An unsupervised Bayesian image-segmentation method produces binary representations that are compared with class-specific geometric templates inside YOLOv2 bounding boxes. This information is used either during inference or as an additional signal during training. A dedicated dataset is constructed by combining GTSDB and GTSRB samples using seamless cloning and controlled image transformations. Experiments cover ten traffic-sign classes, with 3,000 training and 300 test samples. The selected branched architecture reports 0.647 s runtime and 0.680 mAP, compared with 0.6607 s and 0.680 mAP for baseline YOLOv2. Geometric verification during inference increases mAP to 0.713, while the geometric-feature training variant achieves 0.697 mAP with a reported runtime of 0.6608 s.

Figures & tables

Panel of Figure 1 — Figure 1: Schematic architecture of the proposed system. Intermediate outputs provide early decisions, while the parallel segmentation path supplies explicit geometric information. The diagram follows the structure described in the thesis and is not a direct copy of the original thesis figure.

Configuration	Runtime (s)	mAP
Three layers, placement 1	0.6559	0.680
Three layers, placement 2	0.6470	0.680
Four layers	0.6708	0.680
Five layers	0.6760	0.680

Table 1: Reported evaluation of 2YOLOv-B with different numbers and placements of intermediate output layers.

Model	Runtime (s)	mAP
YOLOv2	0.6607	0.680
2YOLOv-B, whole-image branching	0.6470	0.680
2YOLOv-B, cell-wise branching	0.6527	0.680

Table 2: Reported comparison of the YOLOv2 baseline and the two 2YOLOv-B branching strategies.

2YOLOv-B configuration	Runtime (s)	mAP
Without intermediate feature fusion	0.6470	0.680
With feature fusion	0.6491	0.689

Table 3: Reported effect of feature fusion for 2YOLOv-B intermediate outputs.

Panel of Figure 2 — Figure 2: Geometric-feature pipeline reconstructed from the thesis: segmentation, region extraction, comparison with original and inverted templates, and shape-consistency decision.

Item	Reported value
Detection source	GTSDB
Classification source	GTSRB
Generated samples	More than 10,000
Samples used in reported experiments	3,300
Number of classes	10
Training samples per class	300

Table 4: Reported dataset configuration.

Parameter	Reported setting
Training epochs	160
Initial learning rate	$10^{-3}$
Momentum	0.9
YOLO confidence threshold	0.001
NMS threshold	0.5
Weight-save interval	5,000 batches

Table 5: Reported experimental settings for YOLOv2 and Bayesian segmentation.

Model / strategy	Runtime (s)	mAP
YOLOv2	0.6607	0.680
Geometric features at inference	0.6714	0.713
2YOLOv2-O, trained with geometric features	0.6608	0.697

Table 6: Reported performance of YOLOv2 and geometric-feature variants.

Panel of Figure 3 — Figure 3: Representative qualitative results of the proposed traffic sign recognition system on real-world road scenes under different visual conditions. The examples illustrate traffic sign detection and classification using the proposed YOLOv2-based framework and its variants.

Explore similar work

RIECNN: Real-Time Image Enhanced CNN for Traffic Sign Recognition
Oct 10, 2026Reem Abdel-Salam, Rana Mostafa, Ahmed H. Abdel-GawadConvolutional Neural Networks Autonomous Driving Perception
Computer Engineering Department, Cairo University, Giza, Egypt
Traffic Sign Recognition in Autonomous Driving: Dataset, Benchmark, and Field Experiment
Mar 24, 2026Guoyang Zhao, Weiqing Qi, Kai Zhang +7Autonomous Driving Datasets Autonomous Driving Perception
The Hong Kong University of Science and Technology (Guangzhou), Guangzhou 511453, China · Lingnan University, Hong Kong SAR, China · Shenzhen Unity Drive Innovation Technology Co., Ltd., Shenzhen 518063, China +1
Squeeze-Aware-YOLO: Early Channel Recalibration and Multi-Scale Fusion for Real-Time Traffic Sign Detection
Oct 11, 2026Mizbaul Haque Maruf, Syed Samin Sadaf, Syed Tamzid Bakht +2Object Detection Real-Time Object Detection
Dept. of Computer Science and Engineering, Islamic University of Technology, Boardbazar, Gazipur, Bangladesh

Oct 10, 2026J/K move · Enter open · S save

RIECNN: Real-Time Image Enhanced CNN for Traffic Sign Recognition

Reem Abdel-Salam, Rana Mostafa, Ahmed H. Abdel-Gawad

Computer Engineering Department, Cairo University, Giza, Egypt

Convolutional Neural Networks Autonomous Driving Perception

Open paper AskPDF arXiv

Traffic sign recognition plays a crucial role in the development of autonomous cars to reduce the accident rate and promote road safety. It has been a necessity to address traffic signs that are affected significantly by the environment as well as poor real-time performance for deep-learning state-of-the-art algorithms. In this paper, we introduce Real-Time Image Enhanced CNN ({\our}) for Traffic Sign Recognition. {\our} is a real-time, novel approach that tackles multiple, diverse traffic sign datasets, and out-performs the state-of-the-art architectures in terms of recognition rate and execution time. Experiments are conducted using the German Traffic Sign Benchmark (GTSRB), the Belgium Traffic Sign Classification (BTSC), and the Croatian Traffic Sign (rMASTIF) benchmark. Experimental results show that our approach has achieved the highest recognition rate for all Benchmarks, achieving a recognition accuracy of 99.75% for GTSRB, 99.25% for BTSC and 99.55% for rMASTIF. In terms of latency and meeting the real-time constraint, the pre-processing time and inference time together do not exceed 1.3 ms per image. Not only have our proposed approach achieved remarkably high accuracy with real-time performance, but it also demonstrated robustness against traffic sign recognition challenges such as brightness and contrast variations in the environment.

Abstract

Explore similar work

RIECNN: Real-Time Image Enhanced CNN for Traffic Sign Recognition

Traffic Sign Recognition in Autonomous Driving: Dataset, Benchmark, and Field Experiment

Squeeze-Aware-YOLO: Early Channel Recalibration and Multi-Scale Fusion for Real-Time Traffic Sign Detection