Comparison of Model Efficiency for PPE Detection at Construction Sites in Abancay Under High Solar Radiation and Occlusion Conditions

Janneth Huanaco; Manuel Ibarra Cabrera

doi:10.67294/c4m7px76

Authors

Janneth Huanaco Universidad Nacional Micaela Bastidas de Apurimac Author
Dr. Manuel Ibarra-Cabrera Universidad Nacional Micaela Bastidas de Apurimac Author https://orcid.org/0000-0001-6711-4916

DOI:

https://doi.org/10.67294/c4m7px76

Keywords:

Personal Protective Equipment, YOLO, RT-DETR, Solar Radiation, Occlusion

Abstract

Monitoring Personal Protective Equipment (PPE) compliance at construction sites in the Andean region of Apurímac, Peru, faces challenges due to extreme solar radiation (reported UV indices exceeding 11) and frequent physical occlusion. This study compares the performance of three architectures (YOLOv8n, YOLO11n, and RT-DETR-l) trained on the public PPE-CPXSZ v4 dataset (11 classes, including violations), which were additionally subjected to a synthetically degraded subset using light overexposure and simulated occlusion as a controlled approximation of Andean environmental stress. The models were evaluated using mAP50, mAP50-95, precision, recall, and real inference latency on an NVIDIA GPU. Under normal conditions, RT-DETR-l achieves the highest overall precision (92.66% mAP50 and 71.47% mAP50-95) thanks to its Transformer-type attention, with a latency of 34.8 ms per image (28.7 FPS), similar to that of YOLOv8n and YOLO11n (35.4–36.8 ms), despite its larger size (32.0 M parameters compared to 2.6–3.0 M parameters). YOLO11n maintains the lowest computational cost (2.6 M parameters, 5.5 MB), with an mAP50 of 89.21%. Under the adverse subset, all three models suffer a sharp drop in mAP50, but RT-DETR-l retains the highest relative value (46.40% compared to 24.75%–26.89%). It is concluded that when measuring real inference latency, the speed advantage attributed to YOLO diminishes compared to RT-DETR-l; therefore, model selection should prioritize precision and memory consumption over speed.

References

Adebowale, O. J., & Agumba, J. N. (2026). Smart PPE and wearable technology for construction safety management: A scientometric analysis and state-of-the-art review. Archives of Environmental & Occupational Health, 81(5-6), 127-138. https://doi.org/10.1080/19338244.2026.2655851 DOI: https://doi.org/10.1080/19338244.2026.2655851

Al-Khiami, M. I., & ElHadad, M. (2024). Enhancing Construction Site Safety Using AI: The Development of a Custom YOLOv8 Model for PPE Compliance Detection. 2024 European Conference on Computing in Construction. https://doi.org/10.35490/EC3.2024.307 DOI: https://doi.org/10.35490/EC3.2024.307

Barlybayev, A., Milosz, M., Amangeldy, N., Li, G., Razakhova, B., Tazhibay, A., Nazyrova, A., & Lamasheva, Z. (2026). Systematic Review of Computer-Vision Technologies for Personal Protective Equipment Compliance Monitoring. Computers, 15(6), 388. https://doi.org/10.3390/computers15060388 DOI: https://doi.org/10.3390/computers15060388

Cheng, S., Qi, Y., Wu, R., & Guan, Y. (2025). A Multimodal Agent Framework for Construction Scenarios: Accurate Perception, Dynamic Retrieval, and Explainable Hazard Reasoning. Buildings, 15(24), 4439. https://doi.org/10.3390/buildings15244439 DOI: https://doi.org/10.3390/buildings15244439

Çiftçi, M., Türkdamar, M. U., & Öztürk, C. (2024). Leveraging YOLO Models for Safety Equipment Detection on Construction Sites. Journal of Computing Theories and Applications, 1(4), 492-506. https://doi.org/10.62411/jcta.10453 DOI: https://doi.org/10.62411/jcta.10453

Delhi, V. S. K., Sankarlal, R., & Thomas, A. (2020). Detection of Personal Protective Equipment (PPE) Compliance on Construction Site Using Computer Vision-Based Deep Learning Techniques. Frontiers in Built Environment, 6, 136. https://doi.org/10.3389/fbuil.2020.00136 DOI: https://doi.org/10.3389/fbuil.2020.00136

Faisal Mohammed, A., Mohammed, S., Raheman Mohammed, A., & Abdullah Kamran, S. (2025). Ai-Augmented Safety Management in Construction. International Journal of Computer Science and Information Technology, 17(5), 107-124. https://aircconline.com/abstract/ijcsit/v17n5/17525ijcsit07.html DOI: https://doi.org/10.5121/ijcsit.2025.17507

He, L., Zhou, Y., Liu, L., & Ma, J. (2024). Research and Application of YOLOv11-Based Object Segmentation in Intelligent Recognition at Construction Sites. Buildings, 14(12), 3777. https://doi.org/10.3390/buildings14123777 DOI: https://doi.org/10.3390/buildings14123777

Hokama, F. (2025, March 13). Producción de 17 departamentos creció en el cuarto trimestre del 2024, según el INEI: ¿Qué sectores la impulsaron?. Radio Programa del Perú. https://rpp.pe/economia/economia/inei-produccion-de-17-departamentos-crecio-en-el-cuarto-trimestre-de-2024-que-sectores-la-impulsaron-noticia-1621656

Li, F., Chen, Y., Hu, M., Luo, M., & Wang, G. (2023). Helmet-Wearing Tracking Detection Based on StrongSORT. Sensors, 23(3), 1682. https://doi.org/10.3390/s23031682 DOI: https://doi.org/10.3390/s23031682

Li, S., Lv, Y., Li, M., & Wang, Z. (2023). Detection of Protective Apparatus for Municipal Engineering Construction Personnel Based on Improved YOLOv5s. International Journal of Advanced Computer Science and Applications, 14(7). https://doi.org/10.14569/IJACSA.2023.0140741 DOI: https://doi.org/10.14569/IJACSA.2023.0140741

Lin, B. (2024). Yolov8n-asf-dh: an enhanced safety helmet detection method. IEEE Access, 12, 126313-126328. https://doi.org/10.1109/ACCESS.2024.3435453 DOI: https://doi.org/10.1109/ACCESS.2024.3435453

Liu, Z., Sun, C., & Wang, X. (2024). DST-DETR: Image Dehazing RT-DETR for Safety Helmet Detection in Foggy Weather. Sensors, 24(14), 4628. https://doi.org/10.3390/s24144628 DOI: https://doi.org/10.3390/s24144628

Lyu, Y., Yang, X., Guan, A., Wang, J., & Dai, L. (2024). Construction personnel dress code detection based on YOLO framework. CAAI Transactions on Intelligence Technology, 9(3), 709-721. https://doi.org/10.1049/cit2.12312 DOI: https://doi.org/10.1049/cit2.12312

Mohy, A. A., Bassioni, H. A., Elgendi, E. O., & Hassan, T. M. (2025). Deep Learning-Enabled Computer Vision Model for Automated Safety Compliance in Construction Environments. Journal of Information Technology in Construction, 30(1), 1398. https://doi.org/10.36680/j.itcon.2025.057 DOI: https://doi.org/10.36680/j.itcon.2025.057

Priya, S., & Amshakala, K. (2025). An adaptive fall detection system based on ensemble learning using variants of YOLO V8 retinanet and DETR. Scientific Reports, 15(1), 33161. https://doi.org/10.1038/s41598-025-97634-8 DOI: https://doi.org/10.1038/s41598-025-97634-8

Rong, Z., Wang, G., Sun, Q., Sun, J., Li, R., & Xu, Y. (2026). A Lightweight and Efficient Model for Safety Protection Equipment Detection Based on Improved RT-DETR. IEEE Access, 14, 5510-5521. https://doi.org/10.1109/ACCESS.2025.3629848 DOI: https://doi.org/10.1109/ACCESS.2025.3629848

Saeheaw, T. (2025). SC-YOLO: A Real-Time CSP-Based YOLOv11n Variant Optimized with Sophia for Accurate PPE Detection on Construction Sites. Buildings, 15(16), 2854. https://doi.org/10.3390/buildings15162854 DOI: https://doi.org/10.3390/buildings15162854

Saeheaw, T. (2026). Comparative Evaluation of YOLOv11n Neck-Level Modifications for Precast Component and PPE Object Detection in Construction Environments. Buildings, 16(9), 1728. https://doi.org/10.3390/buildings16091728 DOI: https://doi.org/10.3390/buildings16091728

SENAMHI (2026). UV Radiation. Ministerio del Ambiente del Perú. https://www.senamhi.gob.pe/?p=radiacion-uv

Seth, Y., & Sivagami, M. (2025). Enhanced YOLOv8 Object Detection Model for Construction Worker Safety Using Image Transformations. IEEE Access, 13, 10582-10594. https://doi.org/10.1109/ACCESS.2025.3527511 DOI: https://doi.org/10.1109/ACCESS.2025.3527511

Shanti, M. Z., An, B., Yeun, C. Y., Cho, C.-S., Damiani, E., & Kim, T.-Y. (2025). Enhancing Worker Safety at Heights: A Deep Learning Model for Detecting Helmets and Harnesses Using DETR Architecture. IEEE Access, 13, 151788-151802. https://doi.org/10.1109/ACCESS.2025.3603202 DOI: https://doi.org/10.1109/ACCESS.2025.3603202

Suhail Ahmed, S., Sobnath, D., Turcanu, C., & Rehman, I. U. (2025). Seeing Safety: Computer Vision for Real-Time PPE Monitoring in the Middle East Construction Sector. Proceedings of the AAAI Symposium Series, 6(1), 137-139. https://doi.org/10.1609/aaaiss.v6i1.36044 DOI: https://doi.org/10.1609/aaaiss.v6i1.36044

Wang, J., Zhao, D., Jiang, H., Lin, P., & Tang, X. (2026). LHAT-YOLO: Study on intelligent monitoring algorithm for helmets at construction sites. PLOS One, 21(1), e0339993. https://doi.org/10.1371/journal.pone.0339993 DOI: https://doi.org/10.1371/journal.pone.0339993

Wang, S., Kim, H., Yeo, J., & Kim, J. (2026). YOLOv10-based multi-scale variant object detection for multi-category PPE non-compliance monitoring on construction sites. Scientific Reports, 16, article 24591. https://doi.org/10.1038/s41598-026-53052-y DOI: https://doi.org/10.1038/s41598-026-53052-y

Wang, Z., Cai, Z., & Wu, Y. (2023). An improved YOLOX approach for low-light and small object detection: PPE on tunnel construction sites. Journal of Computational Design and Engineering, 10(3), 1158-1175. https://doi.org/10.1093/jcde/qwad042 DOI: https://doi.org/10.1093/jcde/qwad042

World Health Organization. (2003). The UV index sun protection table. https://www.who.int/publications/m/item/the-uv-index-sun-protection-table

Xing, J., Zhan, C., Ma, J., Chao, Z., & Liu, Y. (2025). Lightweight detection model for safe wear at worksites using GPD-YOLOv8 algorithm. Scientific Reports, 15(1), article 1227. https://doi.org/10.1038/s41598-024-83391-7 DOI: https://doi.org/10.1038/s41598-024-83391-7

Yang, G., Hong, X., Sheng, Y., & Sun, L. (2024). YOLO-Helmet: A Novel Algorithm for Detecting Dense Small Safety Helmets in Construction Scenes. IEEE Access, 12, 107170-107180. https://doi.org/10.1109/ACCESS.2024.3435700 DOI: https://doi.org/10.1109/ACCESS.2024.3435700

Yao, G., Liu, L., Yang, Y., & Cai, X. (2026). A Vision-Based Sensing Framework for PPE Detection and Safety Harness Compliance Recognition in High-Formwork Construction Environments Using YOLO-ILB. Sensors, 26(13), 4147. https://doi.org/10.3390/s26134147 DOI: https://doi.org/10.3390/s26134147

Yun, S., & Kim, H. (2025). YOLOv8-SAMURAI: A Hybrid Tracking Framework for Ladder Worker Safety Monitoring in Occlusion Scenarios. Buildings, 15(11), 1836. https://doi.org/10.3390/buildings15111836 DOI: https://doi.org/10.3390/buildings15111836

Zhou, X., Jia, X., Bai, J., Lv, X., Lv, X., & Zhang, G. (2025). Optimizing Helmet Use Detection in Construction Sites via Fuzzy Logic-Based State Tracking. Sensors, 25(20), 6487. https://doi.org/10.3390/s25206487 DOI: https://doi.org/10.3390/s25206487