Marble Surface Anomaly Detection Using Autoencoder Architecture
Marble is a material that is commonly used for building components such as furniture, flooring, countertops, bathrooms, windows in homes etc. Due to the many uses of marble in various aspects, marble surface detection is important for this industry to improve quality and avoid financial problems tha...
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Departement of Electrical Engineering, Faculty of Engineering, Universitas Brawijaya
2024-05-01
|
| Series: | Jurnal EECCIS (Electrics, Electronics, Communications, Controls, Informatics, Systems) |
| Subjects: | |
| Online Access: | https://jurnaleeccis.ub.ac.id/index.php/eeccis/article/view/1685 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Marble is a material that is commonly used for building components such as furniture, flooring, countertops, bathrooms, windows in homes etc. Due to the many uses of marble in various aspects, marble surface detection is important for this industry to improve quality and avoid financial problems that may occur. In previous research, many methods such as wavelet transform, Gabor transform, co-occurrence matrix and artificial neural network were implemented in defect detection (fabric or other tasks). In this study we built a platform that aims to detect anomalies on marble surfaces using Autoencoders architecture, Keras library and Python programming language. To test the model that has been created, a marble surface dataset obtained from kaggle.com, one of the largest dataset provider sites, was used and an accuracy of 89% was obtained. The conclusions of this study include the effectiveness of this method in detecting anomalies, the advantages of the autoencoder architecture compared to other methods, and the potential practical applications of these findings in various fields. By utilizing the autoencoder's ability to reconstruct data, anomaly detection can be performed by comparing the reconstructed results with the original data. The main advantage of this approach lies in its ability to tackle the problem of anomaly detection without the need for class labels |
|---|---|
| ISSN: | 2460-8122 |