Real-Time traffic accident detection system on hybrid data with YOLOv9 and YOLOv11 architectures

Abstract

Traffic accidents have become a significant problem in our country and many other countries. Accidents increase the economic and health costs as well as the loss of life. Therefore, the timely detection of accidents is a very important issue. In this study, an accident detection and precaution system was developed on a hybrid dataset using the originally created YOLOv9 and YOLOv11 models in order to prevent these problems. In the first stage of the study, data was prepared for the training of both models by extracting similar images and labels. Hybrid datasets created with images obtained from different sources containing the "accident" and "non-accident" classes were used. In order to eliminate data imbalance, synthetic images were produced with Generative Adversarial Network (GAN), images were labeled and resized to appropriate sizes, and similar and repetitive ones were cleaned. Hierarchical Feature Attention Layer (HFAM) was added to the YOLOv9 model to better capture features, and Dynamic Context Enrichment Layer (DCEL) layer was added to the YOLOv11 model, which increases its sensitivity to environmental factors. Model performances are evaluated on five different scenarios (5 different datasets) with metrics such as mAP50, mAP50-95, accuracy, precision, sensitivity and F1 score, with hyperparameter optimization and k-fold cross validation. In this study, the performance of the proposed models is compared with classical object detection models such as SSD and Fast R-CNN. As a result, it is seen that both models can successfully detect traffic accidents and have high generalization abilities on different data structures.

Keywords

• YOLO
• Deep Learning
• mAP
• Traffic Accidents
• Real-time

YOLOv9 ve YOLOv11 mimarileriyle hibrit veri üzerinde gerçek zamanlı trafik kazası tespit sistemi

Öz

Trafik kazaları ülkemizde ve birçok ülkede önemli bir sorun haline gelmiştir. Kazalar ekonomik ve sağlık maliyetlerinin yanı sıra can kayıplarını da artırmaktadır. Bu nedenle kazaların zamanında tespiti oldukça önemli bir konudur. Bu çalışmada bu sorunların önüne geçmek amacıyla özgün oluşturulmuş YOLOv9 ve YOLOv11 modelleri kullanılarak hibrit veri seti üzerinde bir kaza tespit ve önlem alma sistemi geliştirilmiştir. Çalışmanın ilk aşamasında benzer görseller ve etiketler çıkarılarak her iki modelin eğitimi için veriler hazırlanmıştır. "Kaza" ve "kaza olmayan" sınıflarını içeren, farklı kaynaklardan elde edilen görüntülerle oluşturulan hibrit veri kümeleri kullanılmıştır. Veri dengesizliğini gidermek için Generative Adversarial Network (GAN) ile sentetik görüntüler üretilmiş, görüntüler etiketlenip uygun boyutlara getirilmiş, benzer ve tekrar edenler temizlenmiştir. YOLOv9 modeline özellikleri daha iyi yakalayabilmesi için Hiyerarşik Özellik Dikkat Katmanı (HFAM), YOLOv11 modeline ise çevresel faktörlere duyarlılığını artıran Dinamik Kontekst Zenginleştirme Katmanı (DCEL) katmanı eklenmiştir. Model performansları, mAP50, mAP50-95, doğruluk, kesinlik, duyarlılık ve F1 skoru gibi metriklerle, hiperparametre optimizasyonu ve k-fold çapraz doğrulama ile beş farklı senaryo (5 farklı veri kümesi) üzerinde değerlendirilmiştir. Bu çalışmada önerilen modellerin performansı SSD ve Fast R-CNN gibi klasik nesne algılama modelleri ile karşılaştırılmıştır. Sonuç olarak, her iki modelin de trafik kazalarını başarılı bir şekilde tespit edebildiği ve farklı veri yapılarında genelleme yeteneklerinin yüksek olduğu görülmüştür.

Anahtar Kelimeler

• YOLO
• Derin Öğrenme
• mAP
• Trafik Kazaları
• Gerçek-Zamanlı

Kaynakça

1. https://www.who.int/publications/i/item/9789240086517
2.    Y. Zhang and Y. Sung, Traffic Accident Detection Using Background Subtraction and CNN Encoder–Transformer Decoder in Video Frames. Mathematics, 11(13), 2884, 2023. https://doi.o rg/10.3390/math11132884.
3.    A. Nusari, I. Ozbek and E. Oral, Automatic Vehicle Accident Detection and Classification from Images: A Comparison of YOLOv9 and YOLO-NAS Algorithms. In 2024 32nd Signal Processing and Communications Applications Conference (SIU), pp. 1-4, Mersin, Türkiye, 2024. https://doi.org/ 10.1109/SIU61531.2024.10600761.
4.    V. Sherimon, A. Ismaeel and J.Joy, An Overview of Different Deep Learning Techniques Used in Road Accident Detection. International Journal of Advanced Computer Science & Applications, 14(11), 1-5, 2023.
5.    Z. Zhou, X. Dong, Z. Li and Y. Yang, Spatio-temporal feature encoding for traffic accident detection in VANET environment. IEEE Transactions on Intelligent Transportation Systems, 23(10), 19772-19781, 2022. https://doi.o rg/10.1109/TITS.2022.3147826.
6.    A. Parsa, A. Movahedi, H. Taghipour and A.K. Mohammadian, Toward safer highways, application of XGBoost and SHAP for real-time accident detection and feature analysis. Accident Analysis & Prevention, 136, 105405, 2020. https://doi.org/ 10.1016/j.aap.2019.105405.
7.    K. Agrawal, J. Choudhary, A. Bhattacharya, S. Tangudu and B. Rajitha, Automatic traffic accident detection system using ResNet and SVM. In 2020 Fifth International Conference on Research in Computational Intelligence and Communication Networks (ICRCICN), pp.71-76, Bangalore, India, 2020. https://doi.org/ 10.1109/ICRCICN50933.20 20.9296156.
8.    W. Hu, X. Xiao, D. Xie, T. Tan, and S. Maybank, Traffic accident prediction using 3-D model-based vehicle tracking. IEEE transactions on vehicular technology, 53(3), 677-694, 2004. https://doi.org/ 10.1109/TVT.2004.825772.

9.    J. Hu, J. Bai, J. Yang and J. Lee, Crash risk prediction using sparse collision data: Granger causal inference and graph convolutional network approaches. Expert Systems with Applications, 259, 125315, 2025. https://doi.org/ 10.1016/j.eswa.2024.125315.
10. P. Gao, B. Shuai, R. Zhang and B. Wang, STCM-GCN: a spatial-temproal prediction method for traffic crashes under road network constraints. Transportmetrica B: Transport Dynamics, 13(1), 2434220, 2025. https://doi.org/10. 1080/2168056 6.2024.2434220.
11. N. Dogru and A. Subasi, Traffic accident detection using random forest classifier. In 2018 15th learning and technology conference (L&T), pp. 40-45, Jeddah, Saudi, 2018. https://doi.org/ 10.1109/LT.2018.8368509.
12. V. Adewopo and N. Elsayed, Smart city transportation: Deep learning ensemble approach for traffic accident detection. IEEE Access, 12(1), 59134-59147, 2024. https://doi.org/ 10.1109/ACCESS.2024.3387972.
13. S. Huang, Y. He and X. Chen, M-YOLO: a nighttime vehicle detection method combining mobilenet v2 and YOLO v3. In Journal of Physics: Conference Series, 1883(1), 1-10, 2021. https://doi.org/10.1088/1742-6596/1883/1/012094.
14. K. D. Alemdar and M.Y. Çodur, A New Approach to Detect Driver Distraction to Ensure Traffic Safety and Prevent Traffic Accidents: Image Processing and MCDM. Sustainability, 16(17), 7642, 2024. https://doi.org/ 10.3390/su16177642.
15. A. Ayad and M.E. Abdulmunim, Detecting abnormal driving behavior using modified DenseNet. Iraqi Journal for Computer Science and Mathematics, 4(3), 48-65, 2023. https://doi.org/ 10.52866/ijcsm.2023.02.03.005.
16. J. Cao, C. Song, S. Song, S. Peng and F. Xiao, Front vehicle detection algorithm for smart car based on improved SSD model. Sensors, 20(16), 4646, 2020. https://doi.org/ 10.3390/s20164646.
17. C. J. Lin and Y. Jhang, Intelligent traffic-monitoring system based on YOLO and convolutional fuzzy neural networks. IEEE Access, 10, 14120-14133, 2022. https://doi.org/ 10.1109/ACCESS.2022.3147866.
18. Y. Ye, S. Li and Y. Zhang, Dynamic multi-graph neural network for traffic flow prediction incorporating traffic accidents. Expert Systems with Applications, 234, 121101, 2023. https://doi.org/10.1016/j.eswa.2023.121101
19. X. Wu and T. Li, A deep learning-based car accident detection approach in video-based traffic surveillance. Journal of Optics, 1-9, 2024. https://doi.org/ 10.1007/s12596-023-01581-4.
20. S. Vinothkumar, S. Varadhaganapathy and K.S. Annamalai, Traffic sign detection using hybrid network of YOLO and Resnet. In 2023 International Conference on Computer Communication and Informatics (ICCCI), pp. 1-7, Coimbatore, India, 2023. https://doi.org/ 10.1109/ICCCI56745.2023.10128337.
21. K. Jaspin, A.A. Bright and M. L. Legin, Accident Detection and Severity Classification System using YOLO Model. In 2024 3rd International Conference on Applied Artificial Intelligence and Computing (ICAAIC), pp. 1160-1167, Salem, India, 2024. https://doi.org/ 10.1109/ICAAIC60222.2024.10575528.
22. A. Manninen, Spatiotemporal Traffic Accident Prediction Using Deep Learning Models.
23. U. Sirisha and S.C.Bolem, Utilizing a hybrid model for human injury severity analysis in traffic accidents. Traitement du Signal, 40(5), 2233, 2024. https://doi.org/ 10.18280/ts.400540.
24. F.M. Talaat, R. M. El-Balka, S. Sweidan, S.A. Gamel and A.M. Al-Zoghby, Smart traffic management system using YOLOv11 for real-time vehicle detection and dynamic flow optimization in smart cities. Neural Computing and Applications, 1-18, 2025. https://doi.org/ 10.1007/s00521-025-11434-9.
25. R. Lupian, C. G.Arong, W. Betinol and D.B. Valdez, 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, Vilnius, Lithuania, 2025. https://doi.org/10.1109/eStream66938.2025.11016862
26. P. Kaladevi, M. Balamurugan, D. Gokul and S. Gopika, Real-Time Traffic Monitoring and Ambulance Prioritization Using YOLOv9 and Deep Learning. In 2025 3rd International Conference on Communication, Security, and Artificial Intelligence (ICCSAI), pp. 2172-2177, Greater Noida, India, 2025. https://doi.org/ 10.1109/ICCSAI64074.2025.11063915.
27. C. Dewi, H. P. Chernovita, S. Philemon and P. Chen, Integration of yolov9 and contrast limited adaptive histogram equalization for nighttime traffic sign detection, Mathematical Modelling of Engineering Problems, 12(1), 37-45, 2025. https://doi.org/ 10.18280/mmep.120105.
28. A. Pangestu and M. Putra, YOLOv9–based Traffic Sign Detection under Varying Lighting Conditions. Jurnal Teknik Informatika (Jutif), 6(1), 291-300, 2025. https://doi.org/ 10.52436/1.jutif.202 5.6.1.3917.
29. https://universe.roboflow.com/car-accident-detection-2uoh6/car-accident-detection-7zjkk
30. https://universe.roboflow.com/ambulance-0rcqn/accident_detection-trmhu/dataset/4
31. https://www.cvlibs.net/datasets/kitti/
32. A. Obukhov and M. Krasnyanskiy, Quality assessment method for GAN based on modified metrics inception score and Fréchet inception distance. In Proceedings of the Computational Methods in Systems and Software, pp. 102-114, 2020.
33. T. Chien and S. Chiang, YOLOv9 for fracture detection in pediatric wrist trauma X‐ray images. Electronics Letters, 60(11), 13248, 2024. https://doi.org/ 10.1049/ell2.13248.
34. N. Jegham, C. Koh and A. Hendawi, Evaluating the evolution of yolo (you only look once) models: A comprehensive benchmark study of yolo11 and its predecessors. arXiv preprint arXiv:2411.00201, 2024. https://doi.org/ 10.48550/arXiv.2411.00201.
35. R. Khanam and M. Hussain, Yolov11: An overview of the key architectural enhancements. arXiv preprint arXiv:2410.17725, 2024. https://doi.org/ 10.48550/arXiv.2410.17725.
36. A. Karaman and D. Karaboga, Robust real-time polyp detection system design based on YOLO algorithms by optimizing activation functions and hyper-parameters with artificial bee colony (ABC). Expert systems with applications, 221, 119741, 2023. https://doi.org/ 10.1016/j.eswa.2023.119741.
37. S.T. Fu, A novel road traffic accidents recognition model for intersections in mixed-traffic environment using deep learning-based scene and feature understanding (Doctoral dissertation, Swinburne), 2025.
38. R. Bhargavi, Y. Nagendra and S. Ali, Real-Time Traffic Accident Detection Using I3d-Convlstm2d and Optical Flow. International Journal of Human Computations & Intelligence, 4(3), 453-464, 2025. https://doi.org/ 10.5281/zenodo.15263530.
39. B. Liu, Design of Multifunctional Integrated Real-Time Traffic Supervision System. In 2025 IEEE 5th International Conference on Power, Electronics and Computer Applications (ICPECA, pp. 400-406, Shenyang, China, 2025. https://doi.org/ 10.1109/ICPECA63937.2025.10928836.
40. C. Dewi and P. Chen, Integration of YOLOv9 and Contrast Limited Adaptive Histogram Equalization for Nighttime Traffic Sign Detection. Mathematical Modelling of Engineering Problems, 12(1), 2025. https://doi.org/ 10.18280/mmep.120105.
41. Z. Shi, Y. Wang, D. Guo, and F. Sun, The Urban Intersection Accident Detection Method Based on the GAN-XGBoost and Shapley Additive Explanations Hybrid Model. Sustainability, 17(2), 453, 2025. https://doi.org/ 10.3390/su17020453.
42. F. Ortataş and E. Çetin, A Novel Solution to the Real-Time Lane Detection and Tracking Problem for Autonomous Vehicles by Using Faster R-CNN and Mask R-CNN. International Journal of Automotive Science And Technology, 9(1), 71-80, 2025. https://doi.org/ 10.30939/ijastech..1563319.
43. G. Karthick and P. Whig, Enhancing Road Safety A Review of Deep Learning Techniques for Accident Avoidance. SGS-Engineering & Sciences, 1(1), 2025.
44. O. Zhang and Y. Fu, Effective traffic density recognition based on ResNet-SSD with feature fusion and attention mechanism in normal intersection scenes. Expert Systems with Applications, 261, 125508, 2025. https://doi.org/ 10.1016/j.eswa.2024.125508