Education, Science, Technology, Innovation and Life
Open Access
Sign In
  1. Home
  2. Journals
  3. Journal of Artificial Intelligence Practice
  4. Vol 9, Issue 1, 2026

An RSPD-ECA-Enhanced YOLO Detector and Vision-Language Collaborative Assessment Framework for Traffic Accident Video Understanding

Download as PDF

DOI: 10.23977/jaip.2026.090115 | Downloads: 1 | Views: 20

Author(s)

Pinxi Zhou 1, Xiaoyu Yu 1

Affiliation(s)

1 Beijing National Day School, Beijing, 100088, China

Corresponding Author

Pinxi Zhou

ABSTRACT

Automated traffic accident video analysis requires reliable object localization and concise semantic reporting under response-time constraints. This paper proposes YOLO-RSPD-ECA, a lightweight detector enhancement for accident scenes, and connects it with a vision-language reporting workflow. RSPDConv replaces selected stride-2 convolution paths with a residual spatial-preserving downsampling design, using a pooling detail branch to reduce local information loss. ECA attention is inserted after backbone and neck feature blocks to strengthen accident-relevant channels with limited parameter overhead. YOLO-RSPD-ECA outputs boxes, labels, and key-frame evidence, which are provided to Qwen3.5-397B-A17B for structured accident reports and then reviewed by domain experts. Experiments on 33,462 accident images and 101 test videos show consistent gains over YOLOv8, YOLO11, and YOLO26 baselines. The YOLOv8 mAP50-95 improves from 0.4438 to 0.4754; all videos are processed successfully, with an average expert correctness score of 0.833 and an acceptable-report rate of 85.15%.

KEYWORDS

Traffic accident detection; YOLO; Vision-language model; Video understanding

CITE THIS PAPER

Pinxi Zhou, Xiaoyu Yu. An RSPD-ECA-Enhanced YOLO Detector and Vision-Language Collaborative Assessment Framework for Traffic Accident Video Understanding. Journal of Artificial Intelligence Practice (2026). Vol. 9, No. 1, 132-141. DOI: http://dx.doi.org/10.23977/jaip.2026.090115.

REFERENCES

[1] World Health Organization. Global status report on road safety 2023. Geneva: World Health Organization, 2023. https://www.who.int/publications/i/item/9789240086517.
[2] Wang K, Li Z. Global, regional, and national burdens of road injuries from 1990 to 2021: Findings from the 2021 Global Burden of Disease Study. Injury, 2025, 56: 112221.
[3] Yao Y, Wang X, Xu M, Pu Z, Wang Y, Atkins E, Crandall D J. DoTA: Unsupervised detection of traffic anomaly in driving videos. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023, 45(1): 444-459.
[4] Pawar K, Jalem R S, Tiwari V. Deep learning based detection and localization of road accidents from traffic surveillance videos. ICT Express, 2022, 8(3): 379-387.
[5] Kumar P P, Kant K. TU-DAT: A computer vision dataset on road traffic anomalies. Sensors, 2025, 25(11): 3259.
[6] Liu H, Hu X, Sun G, Zhang W, Zhan J, Fang H, Li Y, Ma W. Intelligent traffic accident detection system in complex dynamic scenarios based on the dual-stream spatiotemporal-fusion model. Engineering Applications of Artificial Intelligence, 2025, 162(Part E): 112675.
[7] Tan M, Pang R, Le Q V. EfficientDet: Scalable and Efficient Object Detection. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020: 10781-10790.
[8] Carion N, Massa F, Synnaeve G, Usunier N, Kirillov A, Zagoruyko S. End-to-End Object Detection with Transformers. In: Proceedings of the European Conference on Computer Vision (ECCV), 2020: 213-229.
[9] Wang C Y, Bochkovskiy A, Liao H Y M. Scaled-YOLOv4: Scaling Cross Stage Partial Network. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2021: 13029-13038.
[10] Ge Z, Liu S, Wang F, Li Z, Sun J. YOLOX: Exceeding YOLO Series in 2021. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), 2021: 988-999.
[11] Wang C Y, Bochkovskiy A, Liao H Y M. YOLOv7: Trainable bag-of-freebies sets new state-of-the-art for real-time object detectors. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2023: 7464-7475.
[12] Wang A, Chen H, Liu L, Chen K, Lin Z, Han J, Ding G. YOLOv10: Real-Time End-to-End Object Detection. In: Advances in Neural Information Processing Systems (NeurIPS), 2024, 37.
[13] Radford A, Kim J W, Hallacy C, Ramesh A, Goh G, Agarwal S, et al. Learning Transferable Visual Models from Natural Language Supervision. In: Proceedings of the 38th International Conference on Machine Learning (ICML), 2021: 8748-8763.
[14] Li J, Li D, Savarese S, Hoi S. BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models. In: Proceedings of the 40th International Conference on Machine Learning (ICML), 2023: 19730-19742.
[15] Liu H, Li C, Wu Q, Lee Y J. Visual Instruction Tuning. In: Advances in Neural Information Processing Systems (NeurIPS), 2023, 36: 34892-34916.
[16] Abootorabi M M, Zobeiri A, Dehghani M, et al. Ask in Any Modality: A Comprehensive Survey on Multimodal Retrieval-Augmented Generation. In: Findings of the Association for Computational Linguistics: ACL 2025, 2025: 16776-16809.
[17] Wang Q, Wu B, Zhu P, Li P, Zuo W, Hu Q. ECA-Net: Efficient Channel Attention for Deep Convolutional Neural Networks. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020: 11534-11542.
[18] He K, Zhang X, Ren S, Sun J. Deep Residual Learning for Image Recognition. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2016: 770-778.

Subscription
E-Mail Alert
Most popular papers
Call for Special Issue
Publication Ethics & OA Statement
Plagiarism Policy
Contact Us
Downloads: 26628
Visits: 817854

Sponsors, Associates, and Links

All published work is licensed under a Creative Commons Attribution 4.0 International License.

Copyright © 2016 - 2031 Clausius Scientific Press Inc. All Rights Reserved.