Revolutionizing skull classification: leveraging Digital forensics and deep learning in physical Anthropology
Loading...
Files
Date
2025
Journal Title
Journal ISSN
Volume Title
Publisher
NHCE
Abstract
This study explores the integration of digital forensics and deep learning to revolutionize
skull classification within physical anthropology. It employs advanced image processing
techniques and convolutional neural networks (CNNs) for accurate identification and
classification of skulls, particularly based on the presence or absence of the mandible. The
research process begins with preprocessing skull images, including resizing, noise removal,
and contrast enhancement, to ensure high-quality inputs. Feature extraction techniques
such as Gray-Level Co-occurrence Matrix (GLCM), Discrete Wavelet Transform (DWT),
Gabor filters, and Segmentation-based Fractal Texture Analysis (SFTA) are used to capture
intricate textural and structural details of the skulls.
The study compares traditional methods utilizing Support Vector Machines (SVMs) with
the proposed CNN-based approach, demonstrating the latter's superior performance with
significantly enhanced accuracy metrics. Data augmentation techniques like rotation,
flipping, and scaling further bolster model robustness and mitigate overfitting. Evaluation
metrics, including accuracy, precision, recall, and F1 score, validate the methodology's
effectiveness, achieving classification accuracies exceeding 99%.
This innovative framework supports anthropologists, forensic scientists, and
archaeologists in efficient management, analysis, and preservation of anthropological
collections. Its application extends to museum collections, forensic investigations,
archaeological research, and medical studies. The findings emphasize the transformative
potential of integrating machine learning and digital forensics in anthropological research,
offering a robust tool for advancing our understanding of human evolution and cultural
heritage