Convolutional Neural Networks for Detecting Ship Hatch Closing Moment: A Case Study Using YOLO Family

• Francisco Carlos Silva Pimentel UEMA
• Roberto de Pádua Carvalho Reis UEMA
• Daniel Lima Gomes Jr. IFMA
• Siti Sarah Maidinz INTI International University
• Omar Andres Carmona Cortes IFMA / UEMA

DOI: https://doi.org/10.5753/wvc.2023.27532

Resumo

This paper presents a neural network-based computer vision approach for detecting ship cargo-hold hatch closure. The investigation is relevant since weather conditions, especially rainfall, cause damage to cargo such as sodium sulfate, sugar, corn, corn bran, and potassium chloride, among others. Registering when the cargo hold hatch is closed could prevent damage to the cargo, avoiding prejudice to transportation companies. Our proposal uses YOLO framework vision detection as an economical alternative to the current state-of-the-art for detecting ship hatch closing with expensive and complex solutions. This investigation presents an experiment in a tailored dataset, and results are applied to real-time video detection that validates a stable and accurate solution to the problem of ship hatch detection. Results have shown that even though regular YOLO v4 reaches better metrics, with an accuracy of 91.55%, Fast YOLO v4 is better for real-time detection but with a penalty of lower accuracy.

Palavras-chave: Hatch cover closure, hatch recognition, neural networks, YOLO, computer vision

Referências

Li, Y.; Li, Z.; Yang, Y.; Zhao, L. and Yang, L., 2023. A fast recognition algorithm of ship hatch in bulk cargo terminal based on point cloud contour extraction. Measurement and Control, 56(1-2), pp.228-2.

Alzubaidil, L.; J. Zhang; A. J. Humaidi; A. Al-Dujali; Y. Duan; O. Al-Shamma; J. Santamaria; M. A. Fadhel; M. Al-Amidie e L. Farhan (2021), Review of deep learning: concepts, CNN architectures, challenges, applications, future directions, Journal of Big Data 8 (53), pp. 1–74.

Redmon J.; Divvala,S.; Girshick, R and Farhadi, You Only Look Once: Unified, real-time object detection, IEEE CVPR, 2016, Retrieved from: https://arxiv.org/pdf/1506.02640.pdf.

Sharma, M.; MT, M. A.; Ali, N.; Kumar, N.; Gopal, N. and Xavier, N., Single pull macgregor type hatch cover, Project Report, Bachelor of technology, Cochin University of Science and Technology, 2011.

Hyla, P., The crane control systems: A survey, Proceedings of 17th International Conference on Methods and Models in Automation and Robotics (Miedzyzdroje, Poland), 2012, pp. 505-50936.

Kaleci,B. and Turgut, K., Comparison of Deep Learning Techniques For Detection of Doors In Indoor Environments, Eskisehir Osmangazi Üniversitesi Mühendislik ve Mimarlık Fakültesi Dergisi, vol.29, no.3, pp.396, 2021.

Ziang, Z.; Liu, X; Ma, M.; Wu, G. and Farrel, J. A., LiDAR-Based Hatch Localization, Remote Sensing, 14 (20), 2022, 5069.

Miao, Y.; Li, C.; Li, Z.; Yang, Y. and Yu, X., A novel algorithm of ship structure modeling and target identification based on point cloud for automation in bulk cargo terminals. Measurement and control, 54(3-4), 2021, pp.155–163.

Mi, C.; Shen, Y.; Mi, W. and Huang, Y., Ship identification algorithm based on 3D point cloud for automated ship loaders. Journal of Coastal Research, (73), 2015, pp.28-34.

Chen, X.; Yang, Y.; Wang, S.; Wu, H.; Tang, J.; Zhao, J. and Wang, Z., Ship type recognition via a coarse-to-fine cascaded convolution neural network. The Journal of Navigation, 73(4), 2020, pp.813-832.

Zhao, R.; Wang, J.; Zheng, X.; Wen, J.; Rao, L. and Zhao, J., 2020. Maritime visible image classification based on double transfer method. IEEE Access, 8, pp.166335-166346.