Intelligent ROI-Based Vehicle Counting Framework for Automated Traffic Monitoring

Mohamed A. Abdelwahab; Zaynab Al-Ariny; Mahmoud Fakhry; El-Sayed Hasaneen

cs.AIApr 14, 2026

Intelligent ROI-Based Vehicle Counting Framework for Automated Traffic Monitoring

Authors: Mohamed A. Abdelwahab, Zaynab Al-Ariny, Mahmoud Fakhry, El-Sayed Hasaneen

Organizations: Electrical Engineering Dept., Faculty of Energy Engineering, Aswan University, Aswan, Egypt · Information Technology Dept., New Assiut Technological University (NATU), Assiut, Egypt · Electrical Engineering Dept., Faculty of Engineering, Sohag University, Sohag, Egypt · Electrical Engineering Dept., Faculty of Engineering, Aswan University, Aswan, Egypt ·

Abstract

Accurate vehicle counting through video surveillance is crucial for efficient traffic management. However, achieving high counting accuracy while ensuring computational efficiency remains a challenge. To address this, we propose a fully automated, video-based vehicle counting framework designed to optimize both computational efficiency and counting accuracy. Our framework operates in two distinct phases: estimation and prediction. In the estimation phase, the optimal region of interest (ROI) is automatically determined using a novel combination of three models based on detection scores, tracking scores, and vehicle density. This adaptive approach ensures compatibility with any detection and tracking method, enhancing the framework's versatility. In the prediction phase, vehicle counting is efficiently performed within the estimated ROI. We evaluated our framework on benchmark datasets like UA-DETRAC, GRAM, CDnet 2014, and ATON. Results demonstrate exceptional accuracy, with most videos achieving 100% accuracy, while also enhancing computational efficiency, making processing up to four times faster than full-frame processing. The framework outperforms existing techniques, especially in complex multi-road scenarios, demonstrating robustness and superior accuracy. These advancements make it a promising solution for real-time traffic monitoring.

Figures & tables

Figure 1: The optimal region for vehicle counting varies considerably between videos. Interestingly, each video demonstrates strong internal consistency, with the best region-highlighted in yellow-remaining relatively stable throughout the video. (a), (b) and (c) show the counting accuracy of different videos from the UA-DETRAC dataset

Figure 2: Key components of the proposed AIR-VC framework, including estimation and prediction phases

Figure 3: The camera-limited regions (highlighted in red) for different scenarios from the UA-DETRAC dataset. For the area near the camera, the camera angle restricts the visibility of the leftmost and rightmost lanes in (a) and (b). In (c), an obstacle hides the middle area of the frame

Figure 4: Identifying the HDDR: A Heatmap is generated using the previously obtained detection scores. The HDDR (region within $cl_{1}$ and $cl_{2}$ ) is identified according to the generated heat map. The HDDR is represented by the wider region with the most vehicle detections

Figure 5: Selecting the optimal counting line(the red dashed line) within the HDDR which indicated by ( $cl_{1}$ , $cl_{2}$ )

Figure 6: Samples of selected ROI for videos from UA-DETRAC Dataset. Samples show videos with (a) vertical and (b) horizontal traffic flow

Sequence	Entire Frame		AIR-VC		Speed
	$h\times w$	FPS	$h\times w$	FPS	Imp.
M_30	480 $\times$ 800	32	98 $\times$ 800	114	255%
Highway	240 $\times$ 320	26	112 $\times$ 320	53	103%
HighwayII	240 $\times$ 320	26	36 $\times$ 320	115	337%
MVI_63521	540 $\times$ 960	32	99 $\times$ 960	114	258%
MVI_40992	540 $\times$ 960	26	540 $\times$ 376	32	23%

Table 1: Processing speed (FPS) using the entire frame and the proposed AIR-VC for each video sequence.

h\times w

refers to the sequence resolution (Height

\times

Width). The last column shows the speed improvement achieved by using AIR-VC compared to using the entire frame.

Figure 7: Counting accuracy for M-30, Highway and HighwayII video sequences using $cl_{o}$ selected by the proposed AIR-VC (red line) and other uniformly distributed counting lines across the frame (white lines)

Figure 8: Counting accuracy for video sequences from UA-DETRAC dataset using clo selected by the proposed AIR-VC (red line) and other uniformly distributed counting lines across the frame (white lines)

Sequence	M-30	Highway	HighwayII
lanes	4	2	4
S. Li [ 6 ]	100%	96.3%	N/A
Abdelwahab [ 8 ]	93.51%	96.43%	95.83%
Zaynab [ 9 ]	98.7%	100%	97.9%
A. Gomaa [ 2 ]	100%	100%	97.9%
Adson [ 1 ]	98.7%	100%	97.92%

Table 2: Counting accuracy comparison for M-30, Highway and HighwayII video sequences.

Sequence		S. Li [ 7 ]	Yang [ 32 ]	Proposed
name	lanes			AIR-VC
MVI_63521	3+3	97.56%	N/A	100%
MVI_40992	2	N/A	N/A	100%
MVI_40963	5	98.57%	N/A	100%
MVI_40192	4+4	99.08%	97.96%	98.9%

Table 3: Counting accuracy comparison for video sequences from UA-DETRAC dataset.

Sequence		Accuracy
name	lanes
M-30	3+3+4	99.1%
HighwayII	3+4	85.7%
MVI_40963	1+5+3	98.1%

Table 4: Counting accuracy for M-30, HighwayII and MVI_40963 video sequences considering multi-roads scenario using the proposed AIR-VC.

Figure 9: Counting accuracy considering multi-roads scenario using $cl_{o}$ selected by the proposed AIR-VC (red line) and other uniformly distributed counting lines across the frame (white lines)

Object Counting

Object counting estimates the number and locations of target instances in images or video, from dense crowds to fine-grained categories.

See all 61 papers

Intelligent Traffic Monitoring with YOLOv11: A Case Study in Real-Time Vehicle Detection
Apr 5, 2026Shkelqim SherifiObject Counting Video Surveillance
Dept. of Computer Science, University of Tetovo, Tetove, R.N.Macedonia
Attention-Augmented YOLOv8 with Ghost Convolution for Real-Time Vehicle Detection in Intelligent Transportation Systems
Apr 22, 2026Syed Sajid Ullah, Muhammad Zunair Zamir, Ahsan Ishfaq +1Intelligent Transportation Systems Object Detection
School of Energy and Electrical Engineering, Chang’an University, Xi’an, China · School of Information Engineering, Chang’an University, Xi’an, China
Video Detector: A Dual-Phase Vision-Based System for Real-Time Traffic Intersection Control and Intelligent Transportation Analysis
Mar 16, 2026Mustafa Fatih Şen, Halûk Gümüşkaya, Şenol PazarIntelligent Transportation Systems Video Object Detection
Ankageo Co. Ltd., Yıldız Technical University, İkitelli Technopark, 34220 Istanbul, Türkiye · Department of Computer Engineering, Istanbul Arel University, 34537 Istanbul, Türkiye · Department of Electrical and Electronics Engineering, Biruni University, 34015 Istanbul, Türkiye

Apr 5, 2026J/K move · Enter open · S save

Intelligent Traffic Monitoring with YOLOv11: A Case Study in Real-Time Vehicle Detection

Shkelqim Sherifi

Dept. of Computer Science, University of Tetovo, Tetove, R.N.Macedonia

Object Counting Video Surveillance

Open paper AskPDF arXiv

Recent advancements in computer vision, driven by artificial intelligence, have significantly enhanced monitoring systems. One notable application is traffic monitoring, which leverages computer vision alongside deep learning-based object detection and counting. We present an offline, real-time traffic monitoring system that couples a pre-trained YOLOv11 detector with BoT-SORT/ByteTrack for multi-object tracking, implemented in PyTorch/OpenCV and wrapped in a Qt-based desktop UI. The CNN pipeline enables efficient vehicle detection and counting from video streams without cloud dependencies. Across diverse scenes, the system achieves (66.67-95.83%) counting accuracy. Class-wise detection yields high precision (cars: 0.97-1.00; trucks: 1.00) with strong recall (cars: 0.82-1.00; trucks: 0.70-1.00), resulting in F1 scores of (0.90-1.00 for cars and 0.82-1.00 for trucks). While adverse weather conditions may negatively impact this performance, results remain robust in typical conditions. By integrating lightweight models with an accessible, cloud-independent interface, this paper contributes to the modernization and development of future smart cities by showing the capacity of AI-driven traffic monitoring systems.

Abstract

Object Counting

Explore similar work

Intelligent Traffic Monitoring with YOLOv11: A Case Study in Real-Time Vehicle Detection

Attention-Augmented YOLOv8 with Ghost Convolution for Real-Time Vehicle Detection in Intelligent Transportation Systems

Video Detector: A Dual-Phase Vision-Based System for Real-Time Traffic Intersection Control and Intelligent Transportation Analysis