An EfficientNet-Based Deep Learning Framework for Automated Container Damage Detection
Abstract
increasingly important in modern port automation. Traditional manual methods for container damage detection are inefficient and prone to
human bias, making them unsuitable for large-scale maritime logistics. To address these issues, this study proposes an improved EfficientNetbased deep learning framework that integrates adaptive preprocessing, a weighted cross-entropy loss function, and cosine annealing learning
rate optimization. The proposed model effectively enhances the recognition of dents, corrosion, and perforations under complex lighting and
background conditions. Experimental results on a dataset of 3, 713 images demonstrate that the model achieves an accuracy of 95.8%, a recall
of 93.6%, an F1-score of 0.945, and a mean average precision (mAP@0.5) of 0.94, surpassing the baseline EfficientNet-B0 by 5.4%. The
lightweight and efficient design of the framework ensures real-time inference capability, providing a practical and scalable solution for intelligent port inspection and maritime safety management.
Keywords
Full Text:
PDFReferences
[1] Lu, A. J., & Lee, S.-H. (2023). "Real-Time Defect Detection Model in Industrial Environment Based on Lightweight Deep Learning
Network." Electronics, 12(21), 4388. https://doi.org/10.3390/electronics12214388.
[2] He, K., Zhang, X., Ren, S., and Sun, J. (2016). "Deep Residual Learning for Image Recognition, " Proceedings of the IEEE Conference
on Computer Vision and Pattern Recognition (CVPR), pp. 770778.
[3] Huang, G., Liu, Z., Van Der Maaten, L., and Weinberger, K. Q. (2017). "Densely Connected Convolutional Networks, " Proceedings of
the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pp. 47004708.
[4] Li, Y., et al. (2022). "Automatic Detection of Hull Surface Corrosion Using Deep CNNs, " Ocean Engineering, 250, 111031. DOI:
10.1016/j.oceaneng.2022.111031.
[5] Wang, Z., Li, B., Li, W., Niu, S., Miao, W., & Niu, T. (2023). "NAS-ASDet: An Adaptive Design Method for Surface Defect Detection
Network using Neural Architecture Search." arXiv preprint, arXiv:2311.10952. https://arxiv.org/abs/2311.10952
[6] Tan, M., and Le, Q. V. (2019). "EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks, " Proceedings of the 36th
International Conference on Machine Learning (ICML), pp. 61056114.
DOI: http://dx.doi.org/10.70711/aitr.v3i4.8195
Refbacks
- There are currently no refbacks.
