Avaliação das redes YOLO para detecção de danos em coberturas em tempo real visando apoio a gestão de manutenção
DOI:
https://doi.org/10.46421/sbtic.v5i00.7510Palavras-chave:
YOLO, Aprendizado profundo, Gestão da manutenção, Inspeção automatizada, Inteligência ArtificialResumo
A detecção automática de defeitos em telhados e coberturas utilizando drones enfrenta desafios relacionados à precisão, eficiência computacional e viabilidade de implementação em sistemas embarcados. Nesse sentido, com o avanço das arquiteturas de Aprendizado Profundo, como o YOLO, tornou-se possível realizar inspeções prediais de forma mais rápida e precisa, reduzindo custos e riscos associados a métodos tradicionais. Este trabalho teve como objetivo avaliar o desempenho de diferentes versões do YOLO (v5 a v11) na detecção de defeitos em telhados, comparando métricas de precisão, mAP50 e tempo de inferência para identificar o modelo mais adequado para aplicações em tempo real. Nesse contexto, foram utilizadas 7.210 imagens capturadas por drones, pré-processadas e divididas em conjuntos de treinamento, validação e teste. Seis modelos YOLO foram treinados e avaliados, com ajustes de hiperparâmetros para garantir comparabilidade. O YOLOv9 destacou-se com a maior precisão (66,6%) e mAP50 (53,9%), enquanto o YOLOv5 apresentou o menor tempo de inferência (6,6 ms). O estudo demonstra que o YOLOv8 é o modelo mais viável para implementação dispositivos embarcados, oferecendo um equilíbrio ideal entre desempenho e eficiência computacional para inspeções prediais em tempo real no apoio a gestão de manutenção.
Downloads
Referências
AFTAB, Muhammad; CHEN, Chien; CHAU, Chi-Kin; RAHWAN, Talal. Automatic HVAC control with real-time occupancy recognition and simulation-guided model predictive control in low-cost embedded system. Energy and Buildings, v. 154, p. 141-156, 2017. DOI: https://doi.org/10.1016/j.enbuild.2017.07.077.
ANIL KUMAR, Jakkani. Artificial Intelligence and Its Applications in the Field of Internet of Things (IoT). International Journal of Research in Science & Engineering, v. 4, n. 5, p. 49-61, 2024. https://doi.org/10.55529/ijrise.45.49.61.
BADUGE, Shanaka Kristombu et al. Artificial intelligence and smart vision for building and construction 4.0: Machine and deep learning methods and applications. Automation in Construction, v. 141, p. 104440, 2022. DOI: https://doi.org/10.1016/j.autcon.2022.104440.
BOCHKOVSKIY, Alexey; WANG, Chien-Yao; LIAO, Hong-Yuan Mark. YOLOv4: Optimal Speed and Accuracy of Object Detection. arXiv preprint arXiv:2004.10934, 2020. Disponível em: https://arxiv.org/abs/2004.10934. Acesso em: 24 jan. 2025.
ĆOROVIĆ, A.; ILIĆ, V.; ÐURIĆ, S.; MARIJAN, M.; PAVKOVIĆ, B. The real-time detection of traffic participants using YOLO algorithm. In: TELECOMMUNICATIONS FORUM (TELFOR), 26., 2018, Belgrade, Serbia. Anais [...]. Belgrade: IEEE, 2018. p. 1-4. DOI: 10.1109/TELFOR.2018.8611986.
CHUA, W. P.; CHEAH, C. C. Deep-Learning-Based Automated Building Construction Progress Monitoring for Prefabricated Prefinished Volumetric Construction. Sensors, v. 24, n. 21, p. 7074, 2024. https://doi.org/10.3390/s24217074.
DAVE, Jay; PATEL, Dr; RAVAL, Dr. Towards Precise Water Quality Assessment: A Deep Learning Approach with Feature Selection in Smart Monitoring Systems. International Journal of Scientific Research in Computer Science, Engineering and Information Technology, v. 10, p. 100-114, jul. 2024. DOI: https://doi.org/10.32628/CSEIT241045.
GARG, C.; BANSAL, V.; GERA, N.; BANSAL, H.; GOEL, K. A Deep Learning and Image Content Analysis enabled approach for Sustainable Transportation. In: 2024 IEEE International Conference on Electronics, Computing and Communication Technologies (CONECCT). Bangalore, Índia, 2024. p. 1-6. doi: 10.1109/CONECCT62155.2024.10677327.
HALEEM, Muhammad Salman. Advances in Artificial Intelligence, Machine Learning and Deep Learning Applications. Electronics, v. 12, n. 18, p. 3780, 2023. https://doi.org/10.3390/electronics12183780.
JIANG, S.; ZHANG, J. Real-time crack assessment using deep neural networks with wall-climbing unmanned aerial system. Computer-Aided Civil and Infrastructure Engineering, v. 35, p. 549–564, 2020. https://doi.org/10.1111/mice.12519.
JENCY, S.; RAMKUMAR, G. Enhancing Deep Learning Model Performance for Surface Crack Detection. In: 2024 5th International Conference on Smart Electronics and Communication (ICOSEC). Trichy, Índia, 2024. p. 1669-1674. doi: 10.1109/ICOSEC61587.2024.10722142.
KHOOJASTE-SARAKHSI, M.; HAGHIGHI, S. S.; GHOMI, S. M. T. F.; MARCHIORI, E. Deep learning for Alzheimer's disease diagnosis: A survey. Artificial Intelligence in Medicine, v. 130, p. 102332, 2022. https://doi.org/10.1016/j.artmed.2022.102332.
LECUN, Y.; BENGIO, Y.; HINTON, G. Deep learning. Nature, v. 521, p. 436–444, 2015. https://doi.org/10.1038/nature14539.
LIN, T. Y. et al. Microsoft COCO: Common Objects in Context. arXiv preprint arXiv:1405.0312, 2014. Disponível em: http://arxiv.org/abs/1405.0312.
MEHTA, Mihir. Intelligent waste management: Maximize profits and minimize waste using IoT and AI. World Journal of Advanced Research and Reviews, v. 24, n. 01, p. 693–701, 2024. https://doi.org/10.30574/wjarr.2024.24.1.3062.
MELO, R.; COSTA, D. Integrating resilience engineering and UAS technology into construction safety planning and control. Engineering, Construction and Architectural Management, v. 26, 2019.
PEINADO, H. S. et al. Potential application of deep learning and UAS for guardrail safety inspection. In: Proceedings of the 31st Annual Conference of the International Group for Lean Construction (IGLC31). 2023. p. 387–398. doi: 10.24928/2023/0148.
PEREZ, Husein; TAH, Joseph H. M.; MOSAVI, Amir. Deep Learning for Detecting Building Defects Using Convolutional Neural Networks. Sensors, v. 19, n. 16, p. 3556, 2019. https://doi.org/10.3390/s19163556.
REDMON, Joseph; DIVVALA, Santosh; GIRSHICK, Ross; FARHADI, Ali. You Only Look Once: Unified, Real-Time Object Detection. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 2016. p. 779–788.
SANZANA, Mirza Rayana et al. Application of deep learning in facility management and maintenance for heating, ventilation, and air conditioning. Automation in Construction, v. 141, p. 104445, 2022. DOI: https://doi.org/10.1016/j.autcon.2022.104445.
SANTOS, Lara Monalisa Alves dos; ZANONI, Vanda Alice Garcia; BEDIN, Eduardo; PISTORI, Hemerson. Deep learning applied to equipment detection on flat roofs in images captured by UAV. Case Studies in Construction Materials, v. 18, p. e01917, 2023. ISSN 2214-5095.
SCAIFE, Anthony D. Improve predictive maintenance through the application of artificial intelligence: a systematic review. Results in Engineering, v. 17, p. 101645, 2024. Disponível em: https://doi.org/10.1016/j.rineng.2023.101645
SHAH, Deval. Mean Average Precision (mAP) Explained: Everything You Need to Know. V7 Labs, [s.d.]. Disponível em: https://www.v7labs.com/blog/mean-average-precision. Acesso em: mar. 2025.
SHEHAB, M. et al. Machine learning in medical applications: A review of state-of-the-art methods. Computers in Biology and Medicine, v. 145, p. 105458, 2022. https://doi.org/10.1016/j.compbiomed.2022.105458.
V, Viswanatha; R K, Chandana; A.C, Ramachandra. Real time object detection system with YOLO and CNN models: a review. arXiv preprint arXiv:2208.00773, 2022. Disponível em: https://arxiv.org/abs/2208.00773.
YANG, Z. et al. A Real-time Tunnel Surface Inspection System using Edge-AI on Drone. In: 2022 IEEE International Conference on Mechatronics and Automation (ICMA). Guilin, China, 2022. p. 749-754. doi: 10.1109/ICMA54519.2022.9856230.
Yang, W., & Jiachun, Z. (2018, July). Real-time face detection based on YOLO. In 2018 1st IEEE international conference on knowledge innovation and invention (ICKII) (pp. 221-224). IEEE.
ZHANG, Guoping; ZHANG, Qiuyue. Building Engineering Cost Prediction Based On Deep Learning: Model Construction and Real-Time Optimization. Journal of Electrical Systems, v. 20, n. 5s, p. 151-164, 2024.
Downloads
Publicado
Como Citar
Edição
Seção
Licença
Copyright (c) 2025 SIMPÓSIO BRASILEIRO DE TECNOLOGIA DA INFORMAÇÃO E COMUNICAÇÃO NA CONSTRUÇÃO
Este trabalho está licenciado sob uma licença Creative Commons Attribution 4.0 International License.
Adota-se Atribuição 4.0 Internacional (CC BY 4.0). Ver detalhamento em https://creativecommons.org/licenses/by/4.0/deed.pt_BR
