Project Team
Students
Diya Mary Rays
Aerospace Engineering
Abington, University Park
Tamara Estrada-Martinez
Computer Science
Harrisburg, Abington
Faculty Mentors
Dr. Yi Yang
Abington
Electrical Engineering
Project
Project Video
Project Abstract
In East Asian culture, paper is a crucial medium for artistic expression, with a rich tradition of papermaking spanning generations. The conservation of these delicate cultural artifacts presents challenges, particularly in obtaining physical samples without damage. To address this, a non-invasive method has been developed to categorize different paper types using artificial intelligence techniques combined with optical coherence tomography (OCT) imaging. This study focuses on classifying kozo- and hemp-fibered paper from Japan using a modified AI model called AlexNet. The model is trained to adapt the dataset used, which are center-cropped OCT scans of different paper samples in their horizontal, vertical and 45-degree orientations. Out of 1024 images, 512 were used for training, which was processed by center cropping which is taking only the center portion of each to concentrate on each paper’s necessary features. This dataset was split into training, testing, and validation sets (70-15-15%) to evaluate the model’s performance. Training results indicated high accuracy, but challenges remained in consistently predicting correct classes for certain samples during the testing of untrained images. Future research aims to enhance data quality and image preprocessing to improve feature extraction and model performance. This study underscores the potential of integrating deep learning with OCT imaging for the non-invasive conservation of East Asian paper heritage.
Evaluate this Project
Use this form link to provide feedback to the presenters, and add your project evaluation for award(s) consideration.