arxiv: 2604.04080 · v1 · submitted 2026-04-05 · 💻 cs.CV · cs.AI· cs.LG

Recognition: no theorem link

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

Shkelqim Sherifi

Authors on Pith no claims yet

Pith reviewed 2026-05-13 17:23 UTC · model grok-4.3

classification 💻 cs.CV cs.AIcs.LG

keywords traffic monitoringYOLOv11vehicle detectionobject trackingreal-time systemssmart cities

0 comments

The pith

Pre-trained YOLOv11 paired with tracking counts vehicles at 67-96 percent accuracy in real traffic videos

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper establishes that an offline real-time traffic monitoring system using a pre-trained YOLOv11 detector with BoT-SORT or ByteTrack tracking can detect and count vehicles from video streams. It reports counting accuracy between 66.67 and 95.83 percent across diverse scenes, with class-wise precision near 1.0 for cars and trucks and F1 scores of 0.90-1.00 for cars and 0.82-1.00 for trucks. A sympathetic reader would care because the approach runs locally on standard hardware without cloud services, using PyTorch, OpenCV, and a Qt interface. This shows a practical path for AI-based traffic tools that function independently. Performance holds in typical conditions though it can drop under adverse weather.

Core claim

The system couples a pre-trained YOLOv11 detector with BoT-SORT/ByteTrack for multi-object tracking to enable efficient vehicle detection and counting from video streams without cloud dependencies, achieving counting accuracy of 66.67-95.83 percent with class-wise F1 scores of 0.90-1.00 for cars and 0.82-1.00 for trucks.

What carries the argument

The YOLOv11 CNN detector combined with BoT-SORT/ByteTrack multi-object tracking to process video streams locally for vehicle identification and counting.

If this is right

The system operates independently of cloud services for local traffic monitoring.
High precision and recall support reliable classification of cars and trucks in ordinary conditions.
The lightweight local pipeline enables accessible deployment for smart city traffic applications.
Results demonstrate real-time capability on standard hardware using PyTorch and OpenCV.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The desktop UI could support quick rollout by traffic departments without specialized servers.
Extending the pipeline to additional classes such as bicycles or pedestrians would broaden its monitoring uses.
Pairing the counts with signal timing data might enable active traffic flow adjustments as a next application.

Load-bearing premise

The pre-trained YOLOv11 model generalizes sufficiently to the tested traffic scenes without scene-specific fine-tuning, and adverse weather is the primary performance limiter.

What would settle it

Measurements on a new set of traffic videos recorded in heavy rain, snow, or from unseen camera angles that show counting accuracy falling below 60 percent would falsify the reported generalization.

read the original abstract

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.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

This is a routine case study applying YOLOv11 to traffic monitoring with a UI, but lacks dataset details and comparisons.

read the letter

The punchline is that this paper is a routine application of YOLOv11 to traffic video analysis, complete with tracking and a desktop interface, but it brings no new algorithms or theoretical advances. They do a solid job putting together a working system. The pre-trained detector paired with BoT-SORT gives counting accuracies between 66.67 and 95.83 percent on their scenes. Precision for cars and trucks is high, recall is strong, and F1 scores are in the 0.82 to 1.00 range. Wrapping it in a Qt UI makes it accessible for offline use, which fits the goal of cloud-independent smart city tools. The main issues are the limited evaluation. Dataset details are missing, including video sources, lengths, and annotation procedures. There's no comparison to prior YOLO versions or other trackers, and no experiments on fine-tuning or handling adverse weather quantitatively. The mention of weather impact stays qualitative. These gaps make the results harder to interpret or build upon. These are common in case studies but limit how much we can learn from the numbers. This work suits readers who want a practical example for deploying vehicle detection in traffic monitoring. It could serve as a template for engineers or students starting similar projects. For a serious referee, I would not recommend it. The contribution is too incremental for peer review in a research venue; it belongs more as an arXiv technical report or a demo at an applied conference. If the code is released, that would increase its usefulness for others trying to replicate the setup.

Referee Report

1 major / 1 minor

Summary. The manuscript presents an offline real-time traffic monitoring system that couples a pre-trained YOLOv11 detector with BoT-SORT/ByteTrack multi-object tracking, implemented in PyTorch/OpenCV and wrapped in a Qt desktop UI. It reports vehicle counting accuracies of 66.67-95.83% across diverse scenes, along with class-wise metrics (cars: precision 0.97-1.00, recall 0.82-1.00, F1 0.90-1.00; trucks: precision 1.00, recall 0.70-1.00, F1 0.82-1.00), claiming robustness in typical conditions while noting potential degradation in adverse weather.

Significance. If the reported metrics are reproducible on the tested scenes, the work provides a practical demonstration of deploying recent YOLO models for cloud-independent traffic monitoring via a lightweight desktop interface. This adds value as an engineering case study for smart-city applications, showing accessible integration of detection and tracking pipelines without novel architectural contributions.

major comments (1)

[Results/Evaluation] Results/Evaluation section: The reported counting accuracy range (66.67-95.83%) and class-wise precision/recall/F1 scores are presented without any description of the test dataset (number of videos, scenes, total frames, or ground-truth annotation process). This detail is load-bearing for assessing the reliability and scope of the central performance claims.

minor comments (1)

[Abstract] The abstract states that 'adverse weather conditions may negatively impact this performance' but provides no quantitative ablation, example frames, or error breakdown by condition to support the statement.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their positive recommendation and constructive comment. We address the major comment below and have made the corresponding revision to the manuscript.

read point-by-point responses

Referee: [Results/Evaluation] Results/Evaluation section: The reported counting accuracy range (66.67-95.83%) and class-wise precision/recall/F1 scores are presented without any description of the test dataset (number of videos, scenes, total frames, or ground-truth annotation process). This detail is load-bearing for assessing the reliability and scope of the central performance claims.

Authors: Thank you for highlighting this important point. We have revised the Results/Evaluation section to provide a comprehensive description of the test dataset, including the number of videos, the specific scenes, the total frames processed, and the process used for generating ground-truth annotations. These details have been added to allow readers to better assess the reliability of our performance claims. revision: yes

Circularity Check

0 steps flagged

No significant circularity

full rationale

The paper is a purely empirical case study that applies a pre-trained YOLOv11 detector plus BoT-SORT/ByteTrack tracking to a set of traffic videos and reports the observed counting accuracy and class-wise detection metrics on those specific scenes. No equations, parameter fitting, theoretical derivations, or uniqueness claims appear anywhere in the manuscript; every reported number is a direct empirical outcome of running the fixed pipeline on the authors' test data. Because the central claims are scoped to the tested implementation and scenes with no generalization or predictive step that could reduce to the inputs by construction, the work is self-contained and exhibits no circularity.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on the assumption that a publicly available pre-trained YOLOv11 checkpoint transfers adequately to the authors' traffic videos without retraining. No new parameters are fitted and no new entities are postulated.

axioms (1)

domain assumption Pre-trained YOLOv11 generalizes to the tested traffic scenes without fine-tuning
The system description states use of a pre-trained detector with no mention of additional training or adaptation steps.

pith-pipeline@v0.9.0 · 5527 in / 1363 out tokens · 44986 ms · 2026-05-13T17:23:38.841591+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

28 extracted references · 28 canonical work pages · 1 internal anchor

[1]

Bajrami, E., & Halili, F. (2024). Exploring The Impact Of Artificial Intelligence On The IoT And Digital Agenda In The Western Balkans: Integrating A Proposed Web Application For Regional Advancement. JNSM Journal of Natural Sciences and Mathematics of UT, 9(17-18), 244-258

work page 2024
[2]

(2025, March)

Chi, B. (2025, March). TSM-YOLO: An Optimized YOLOv11 Model for Traffic State Surveillance. In 2025 7th International Conference on Software Engineering and Computer Science (CSECS) (pp. 1-6). IEEE

work page 2025
[3]

Mitra, R., Sourav, M. A. A., Kim, S., Gulmezoglu, B., & Ceylan, H. (2025). Comparative Case Study: Traffic Monitoring Using YOLOv11-Based Object Detection and Two Tracking Algorithms with Small Uncrewed Aerial Systems. In International Conference on Transportation and Development 2025 (pp. 311-321)

work page 2025
[4]

Tan, S., Pan, W., Deng, L., Zuo, Q., & Zheng, Y. (2025). FF-YOLO: An Improved YOLO11-Based Fatigue Detection Algorithm for Air Traffic Controllers. Applied Sciences, 15(13), 7503

work page 2025
[5]

R., Jin, P., & Adu-Gyamfi, Y

Mandal, V., Mussah, A. R., Jin, P., & Adu-Gyamfi, Y. (2020). Artificial intelligence-enabled traffic monitoring system. Sustainability, 12(21), 9177

work page 2020
[6]

Majumder, D. D. (1988). Computer Vision and Knowledge-Based Computer Systems. IETE Journal of Research, 34(3), 230-245

work page 1988
[7]

& Shoshi, L

Halili, F., Alihajdaraj, E., Berisha, N. & Shoshi, L. (2019). Artificial Intelligence In Work Process Automation

work page 2019
[8]

[Online]

COCO, 2025. [Online]. Available: https://cocodataset.org/#home, (Accessed: 10 April 2025)

work page 2025
[9]

[Online]

Test Videos, 2025. [Online]. Available: https://drive.google.com/drive/folders/1aZZqMP9EWoQ- XYmMbMDHQeUv7l8_Z_5G?usp=sharing

work page 2025
[10]

Fetzer, J. H. (1990). What is artificial intelligence?. In Artificial Intelligence: Its scope and limits (pp. 3-27). Dordrecht: Springer Netherlands

work page 1990
[11]

Gates, E. (1989). Webster’s New World Dictionary. English Today, 5(2), 52-54

work page 1989
[12]

Lupian, R. R. F., Arong, C. G., Betinol, W. S., & Valdez, D. B. (2025, April). Intelligent Traffic Monitoring And Accident Detection System Using YOLOv11 And Image Processing. In 2025 IEEE Open Conference of Electrical, Electronic, and Information Sciences (eStream) (pp. 1-5)

work page 2025
[13]

Spahija, L., & Halili, F. (2017). Review of artificial intelligence development, its impact, and its challenges

work page 2017
[14]

G., & Tenenbaum, J

Barrow, H. G., & Tenenbaum, J. M. (1981). Computational vision. Proceedings of the IEEE, 69(5), 572-595

work page 1981
[15]

In International Journal of Computer Science and Mobile Computing, vol

Festim Halili, Avni Rustemi, Predictive Modeling: Data Mining Regression Technique Applied in a Prototype. In International Journal of Computer Science and Mobile Computing, vol. 5, issue. 8, 2015

work page 2015
[16]

YOLOv4: Optimal Speed and Accuracy of Object Detection

Bochkovskiy, A., Wang, C. Y., & Liao, H. Y. M. (2020). Yolov4: Optimal speed and accuracy of object detection. arXiv preprint arXiv:2004.10934

work page internal anchor Pith review Pith/arXiv arXiv 2020
[17]

(2024, October)

Bakirci, M., Dmytrovych, P., Bayraktar, I., & Anatoliyovych, O. (2024, October). Multi-class vehicle detection and classification with YOLO11 on UAV-captured aerial imagery. In 2024 IEEE 7th International Conference on Actual Problems of Unmanned Aerial Vehicles Development (APUAVD) (pp. 191-196). IEEE

work page 2024
[18]

(2024, December)

Zhou, Z. (2024, December). Traffic accident detection based on YOLOv11. In 2024, IEEE 2nd International Conference on Electrical, Automation and Computer Engineering (ICEACE) (pp. 363-369). IEEE

work page 2024
[19]

K., & Mishra, D

Dash, S., Padhy, S., Das, S. K., & Mishra, D. (2025, May). A Comparative Deep Learning Approach for Vehicle Speed Monitoring using YoLov11. In 2025 International Conference on Intelligent and Cloud Computing (ICoICC) (pp. 1-6). IEEE

work page 2025
[20]

Alif, M. A. R. (2024). Yolov11 for vehicle detection: Advancements, performance, and applications in intelligent transportation systems. arXiv preprint arXiv:2410.22898

work page arXiv 2024
[21]

M., El-Balka, R

Talaat, F. M., El-Balka, R. M., Sweidan, S., Gamel, S. A., & Al- Zoghby, A. M. (2025). Smart traffic management system using YOLOv11 for real-time vehicle detection and dynamic flow optimization in smart cities. Neural Computing and Applications, 1-18

work page 2025
[22]

Pudaruth, S., & Boodhun, I. M. (2024). Reducing Traffic Congestion Using Real-Time Traffic Monitoring with YOLOv8. International Journal of Advanced Computer Science & Applications, 15(10)

work page 2024
[23]

(2024, November)

Borse, R., Bhattacharyya, A., Sarkar, A., & Bhattacharjee, S. (2024, November). Employing the YOLO model for traffic monitoring on roadways. In the 2024 International Conference on Intelligent Computing and Sustainable Innovations in Technology (IC-SIT) (pp. 1-6). IEEE

work page 2024
[24]

A., & Singh, D

Saklani, S., Dhondiyal, S. A., & Singh, D. (2025, April). Real-Time Traffic Management System Using YOLOv8: An analysis of various YOLO Models. In 2025 4th OPJU International Technology Conference (OTCON) on Smart Computing for Innovation and Advancement in Industry 5.0 (pp. 1-6). IEEE

work page 2025
[25]

Y., Maire, M., Belongie, S., Hays, J., Perona, P., Ramanan, D., & Zitnick, C

Lin, T. Y., Maire, M., Belongie, S., Hays, J., Perona, P., Ramanan, D., & Zitnick, C. L. (2014, September). Microsoft Coco: Common objects in context. In European Conference on Computer Vision (pp. 740-755). Cham: Springer International Publishing

work page 2014
[26]

Milan, A., Leal-Taixé, L., Reid, I., Roth, S., & Schindler, K. (2016). MOT16: A benchmark for multi-object tracking. arXiv preprint arXiv:1603.00831

work page Pith review arXiv 2016
[27]

Regulation, P. (2016). Regulation (EU) 2016/679 of the European Parliament and of the Council. Regulation (EU), 679(2016), 10-13

work page 2016
[28]

[Online]

The impact of the General Data Protection Regulation (GDPR) on artificial intelligence, 2020. [Online]. Available: https://www.europarl.europa.eu/RegData/etudes/STUD/2020/641530/ EPRS_STU(2020)641530_EN.pdf, (Accessed: 24 July 2025)

work page 2020