Segmentasi Citra Pengelasan Kapal Menggunakan Convolutional Neural Network

Thomas Brian; Anggarjuna Puncak Pujiputra; Putri Nur Rahayu; Immanuel Freddy Augustino; Parman Parman; Iskia Ipan Dua’ Bone

doi:10.55681/sentri.v4i10.4767

Authors

Thomas Brian Politeknik Perkapalan Negeri Surabaya
Anggarjuna Puncak Pujiputra Politeknik Perkapalan Negeri Surabaya
Putri Nur Rahayu Politeknik Perkapalan Negeri Surabaya
Immanuel Freddy Augustino Politeknik Perkapalan Negeri Surabaya
Parman Parman Politeknik Perkapalan Negeri Surabaya
Iskia Ipan Dua’ Bone Politeknik Perkapalan Negeri Surabaya

DOI:

https://doi.org/10.55681/sentri.v4i10.4767

Keywords:

CNN, Image Segmentation, Nested Unet, Welding Spot

Abstract

Welding inspection plays a critical role in the shipbuilding industry to ensure the integrity and quality of welded joints. However, the prevailing manual inspection procedures are inherently subjective, prone to bias, and result in inconsistent quality assessments. Therefore, there is a strong need for an automated and intelligent system capable of objectively detecting welding points. To address this, we propose an advanced segmentation model based on deep learning and computer vision techniques, specifically utilizing the enhanced Nested UNet architecture with extensive architectural modifications and comprehensive hyperparameter tuning. To further optimize the segmentation performance, we systematically compare different convolutional blocks integrated into the network architecture. The dataset used consists of 548 welding images. Each image is manually annotated using the VGG Image Annotator (VIA) application by marking the weld point areas as polygons. This research focuses on the development of a Nested UNet model, a deep learning-based image segmentation model for detecting weld points from previous models using the UNet architecture. During the training process, performance on both the training and validation datasets is continuously monitored and recorded. This results in several logs recording the training loss, validation loss, training IoU, and validation IoU for each of the three types of convolutional blocks used in the dense bottleneck. Our experimental evaluation shows that the use of VGG - Dense - VGG convolutional blocks in Nested UNet yields the highest performance, achieving a Training Dice score of 0.92970 and a Validation Dice score of 0.89695 on our collected dataset.

Downloads

Download data is not yet available.

References

T. W. Purnomo, F. Danitasari, and D. Handoko, “Weld Defect Detection and Classification based on Deep Learning Method: A Review”, Journal of Computer Science and Information, Vol. 16, No. 1, pp. 77-87, 2023.

A. I. Gavrilov and M. T. Do, “Classification of Weld Defects based on Convolution Neural Network”, Instrument Engineering, No. 2, pp. 23-36, 2021.

S. Surono, M. Rivaldi, D. A. Dewi, and N. Irsalinda, “New Approach to Image Segmentation: U-Net Convolutional Network for Multiresolution CT Image Lung Segmentation”, Emerging Science Journal, Vol. 7, No. 2, pp. 498-506, 2023.

P. Asri, Y. Widiarti, E. P. Purwanti, E. Wismawati, and M. F. T. Arifin, “Obstacles Detection in Underwater Environment Using ROV Based on Convolutional Neural Network”, Indonesian Journal of Computing and Cybernetics Systems, Vol. 19, No. 1, pp. 105-116, 2024.

E. Akanksha and P. R. K. Rao, “A Feature Extraction Approach for Multi-Object Detection Using HoG and LTP”, Intelligent Networks and Systems Society, Vol. 14, No. 5, pp. 259-268, 2021.

A. N. Azhari and Wahyono, “Automatic Detection of Helmets on Motorcyclists Using Faster -RCNN”, Indonesian Journal of Computing and Cybernetics Systems, Vol. 16, No. 4, pp. 357-366, 2022.

M. Ansari, A. Choudhary, and S. Bhosale, “Semantic Segmentation using Convolutional Neural Networks”, International Journal of Engineering Research in Computer Science and Engineering, Vol. 10, No. 6, pp. 31-34, 2023.

L. Yang, H. Wang, W. Meng, and H. Pan, “CvT-UNet: A weld pool segmentation method integrating a CNN and a transformer”, Heliyon, Vol. 10, 2024.

M. Vadduri and P. Kuppusamy, “DAUCD: Deep Attention U-Net for Cataract Detection Leveraging CNN Frameworks”, Intelligent Networks and Systems Society, Vol. 18, No. 1, pp. 184-197, 2025.

H. K. Abass, M. H. Al-hashimi, A. S. Al-Zubaidi, and M. Al-Mukhtar, “GastroFPN Advanced Deep Segmentation Model for Gastrointestinal Disease with Enhanced Feature Pyramid Network Decoder”, Intelligent Networks and Systems Society, Vol. 17, No. 4, pp. 438-450, 2024.

T. A. Putri, T. Suratno, and U. Khaira, “Identification of Incung Characters (Kerinci) to Latin Characters UsingConvolutional Neural Network”, Indonesian Journal of Computing and Cybernetics Systems, Vol. 16, No. 2, pp. 205-216, 2022.

S. Kumaresan, K. S. Jai Aultrin, S. S. Kumar, and M. D. Anand, “Deep learning-based weld defect classification using VGG16 transfer learning adaptive fine-tuning”, International Journal on Interactive Design and Manufacturing, Vol. 17, pp. 2999–3010, 2023.

S. Oh, M. Jung, C. Lim, and S. Shin, “Automatic Detection of Welding Defects Using Faster R-CNN”, Applied Sciences, Vol. 10, No. 23, pp. 8629, 2020.

O. Natan, D. U. K. Putri, and A. Dharmawan, “Deep Learning-based Weld Spot Segmentation using Modified UNet with Various Convolutional Blocks”, In: ICIC Express Letter B: Applications, Vol. 12, No. 12, pp. 1169-1176, 2021.

Y. Yang, S. Dasmahapatra, and S. Mahmoodi, “ADS_UNet: A Nested UNet for Histopathology Image Segmentation”, Expert Systems with Applications, Vol. 226, 2023.

C. Mathews and A. Mohamed, “Nested U-Net with Enhanced Attention Gate and Compound Loss for Semantic Segmentation of Brain Tumor from Multimodal MRI”, Intelligent Networks and Systems Society, Vol. 15, No. 4, pp. 248-256, 2024.

M. Anand and M. A. Sundaram, “Channel and Spatial Attention Aware UNet Architecture for Segmentation of Blood Vessels, Exudates and Microaneurysms in Diabetic Retinopathy”, Intelligent Networks and Systems Society, Vol. 17, No. 2, pp. 1-16, 2024.

B. K. Triwijoyo, A. Adil, and A. Anggrawan, “Convolutional Neural Network with Batch Normalization for Classification of Emotional Expressions Based on Facial Images”, Matrik: Jurnal Manajemen, Teknik Informatika, dan Rekayasa Komputer, Vol. 21, No. 1, pp. 197-204, 2021.

Herman, A. P. Putri, M. N. M. M. Noor, H. Darwis, L. N. Hayati, Irawati, and I. As’ad, “A Comperative Study on Efficacy of CNN VGG-16, DenseNet121, ResNet50V2, And EfficientNetB0 in Toraja Carving Classification”, Indonesian Journal of Data and Science, Vol. 6 No. 1, pp. 123-133, 2025.

T. Barakbayeva and M. Fatih Demirci, “Fully automatic CNN design with inception blocks”, In: Thirteenth International Conference on Digital Image Processing, Singapore, 2021.

W. W. Kusuma, R. R. Isnanto, and A. Fauzi, “Comparison Analysis of CNN Model VGG16 and DenseNet121 using TensorFlow Framework to Detect Animal Images”, Jurnal Teknik Komputer, Vol. 1, No. 4, pp. 141-147, 2023.

A. U. Ruby, P. Theerthagiri, I. J. Jacob, and Y. Vamsidhar, “Binary cross entropy with deep learning technique for Image classification”, International Journal of Advanced Trends in Computer Science and Engineering, Vol. 9, No. 4, pp. 5393-5397, 2020.

Q. Ming and X. Xiao, “Towards Accurate Medical Image Segmentation With Gradient-Optimized Dice Loss”, IEEE Signal Processing Letters, Vol. 31, pp. 191-195, 2024.

H. Tessier, V. Gripon, M. Léonardon, M. Arzel, T. Hannagan, and D. Bertrand, “Rethinking Weight Decay for Efficient Neural Network Pruning”, Journal of Imaging, Vol. 8, pp. 64, 2022.

A. Buslaev, V. I. Iglovikov, E. Khvedchenya, A. Parinov, M. Druzhinin, and A. A. Kalinin, “Albumentations: Fast and Flexible Image Augmentations”, Information, Vol. 11, No. 2, pp. 125, 2020.

S. Prongnuch and S. Sitjongsataporn, “Stability and Steady-State Performance of Hammerstein Spline Adaptive Filter Based on Stochastic Gradient Algorithm”, Intelligent Networks and Systems Society, Vol. 13, No. 3, pp. 112-123, 2020.