BR-RTSD40k: Scaling Brazilian Traffic Sign Recognition via Semi-Supervised Annotation and Transformer-Based Baselines

• Arthur Maciel UFRN
• Davi Almeida UFRN
• Gabriel Dantas UFRN
• Victor Gomes UFRN
• Artur R. Pinto UFRN
• Maria Klara B. Sena UERN
• Valério de Medeiros Júnior IFRN
• Frederico Lopes UFRN
• Daniel S. A. Araújo UFRN

DOI: https://doi.org/10.5753/eniac.2026.31471

Resumo

Traffic sign recognition is a fundamental component of intelligent transportation systems and autonomous driving applications. Although several public traffic sign datasets have been proposed in recent years, there remains a lack of large-scale datasets representing the diversity of Brazilian traffic signage and road conditions. This limitation hinders the development and evaluation of robust models tailored to the Brazilian context. This paper presents the construction of a Brazilian traffic sign dataset, created from street-level imagery collected in real-world driving scenarios and complemented with samples from publicly available sources. To enable efficient large-scale annotation, we propose a semi-supervised labeling pipeline that combines automatic object detection with manual validation and refinement. The resulting dataset comprises 41,016 annotated images, specifically curated to feature the Brazilian R-type (regulatory) signs targeted in this study. To establish a benchmark for future research, we trained an RT-DETR-based object detector and a Convolutional Neural Network (CNN) for sign classification, both of which were rigorously evaluated. Experimental results demonstrate the feasibility of the proposed dataset and provide strong baseline performance for Brazilian traffic sign recognition tasks. To foster reproducible research and accelerate advances in intelligent transportation systems, the complete dataset will be publicly released to the research community.

Palavras-chave: Traffic Sign, RTDETR, Dataset, Semi-Supervised Annotation

Referências

Carion, N., Massa, F., Synnaeve, G., Usunier, N., Kirillov, A., and Zagoruyko, S. (2020). End-to-end object detection with transformers. In ECCV, page 213–229, Berlin, Heidelberg. Springer-Verlag.

Chen, H., Ali, M. A., Nukman, Y., Razak, B. A., Turaev, S., Chen, Y., Zhang, S., Huang, Z., Wang, Z., and Abdulghafor, R. (2024). Computational methods for automatic traffic signs recognition in autonomous driving on road: A systematic review. Results in Engineering, 24:103553.

Dalborgo, V., Murari, T. B., Madureira, V. S., Moraes, J. G. L., Bezerra, V. M. O. S., Santos, F. Q., Silva, A., and Monteiro, R. L. S. (2023). Traffic sign recognition with deep learning: Vegetation occlusion detection in brazilian environments. Sensors, 23(13).

Dilek, E. and Dener, M. (2023). Computer vision applications in intelligent transportation systems: A survey. Sensors, 23(6).

Ertler, C., Mislej, J., Ollmann, T., Porzi, L., and Kuang, Y. (2020). Traffic sign detection and classification around the world. In ECCV.

Houben, S., Stallkamp, J., Salmen, J., Schlipsing, M., and Igel, C. (2013). Detection of traffic signs in real-world images: The German Traffic Sign Detection Benchmark. In International Joint Conference on Neural Networks, number 1288.

Lin, T.-Y., Maire, M., Belongie, S., Hays, J., Perona, P., Ramanan, D., Dollár, P., and Zitnick, C. L. (2014). Microsoft coco: Common objects in context. In ECCV, pages 740–755. Springer International Publishing.

Mapillary (2026). Mapillary API v4 documentation. [link]. visited on 2026-06-21.

Nellore, K. and Hancke, G. P. (2016). A survey on urban traffic management system using wireless sensor networks. Sensors.

Projetos (2023). Brazilian traffic signs dataset. [link]. visited on 2026-06-21.

Redmon, J., Divvala, S., Girshick, R., and Farhadi, A. (2016). You only look once: Unified, real-time object detection. In CVPR.

Silva, R., Prado, B., Matos, L. N., Santo, F., Zanchettin, C., and Novais, P. (2021). Construction of brazilian regulatory traffic sign recognition dataset. In Tavares, J. M. R. S., Papa, J. P., and González Hidalgo, M., editors, Progress in Pattern Recognition, Image Analysis, Computer Vision, and Applications.

Stallkamp, J., Schlipsing, M., Salmen, J., and Igel, C. (2012). Man vs. computer: Benchmarking machine learning algorithms for traffic sign recognition. Neural Networks.

Zhao, Y., Lv, W., Xu, S., Wei, J., Wang, G., Dang, Q., Liu, Y., and Chen, J. (2024). Detrs beat yolos on real-time object detection.

Zhu, Z., Liang, D., Zhang, S., Huang, X., Li, B., and Hu, S. (2016). Traffic-sign detection and classification in the wild. In CVPR.