A Hybrid Capsule Network-based Deep Learning Framework for Deciphering Ancient Scripts with Scarce Annotations: A Case Study on Phoenician Epigraphy

Rizk, Rodrigue; Rizk, Dominick; Rizk, Frederic; Kumar, Ashok

doi:10.1109/mwscas47672.2021.9531798

Cited by 9 publications

(4 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Haliassos et al [23] detected Egyptian hieroglyphs in papyri using deep CNNs. Rizk et al [24] designed a capsule network architecture for classifying Phoenician letters. Moustafa et al [25] built an end-to-end hieroglyph translation system based on computer vision.…”

Section: Machine Learning For Ancient Scriptsmentioning

confidence: 99%

“…However, the documented corpora cover only limited linguistic and temporal diversity subsets. Many are limited to less than 50,000 examples [24,35,36]. These datasets also suffer from metadata inconsistencies, class imbalance, unrealistic data, and inadequate diversity.…”

Section: Training Datasets For Ancient Scriptsmentioning

confidence: 99%

“…Focusing only on typographic features fails to encapsulate damage and material wear. Rizk et al [24] introduced a framework using synthetic data generation and augmentation to train a neural network for classifying Phoenician script letters (see Fig 3). However, their pipeline relied on creating synthetic letters as white glyphs on black backgrounds rather than realworld images.…”

Section: Synthetic Data For Ancient Scriptsmentioning

confidence: 99%

See 2 more Smart Citations

Deep Aramaic: Towards a synthetic data paradigm enabling machine learning in epigraphy

Aioanei,

Hunziker-Rodewald,

Klein

et al. 2024

PLoS ONE

View full text Add to dashboard Cite

Epigraphy is witnessing a growing integration of artificial intelligence, notably through its subfield of machine learning (ML), especially in tasks like extracting insights from ancient inscriptions. However, scarce labeled data for training ML algorithms severely limits current techniques, especially for ancient scripts like Old Aramaic. Our research pioneers an innovative methodology for generating synthetic training data tailored to Old Aramaic letters. Our pipeline synthesizes photo-realistic Aramaic letter datasets, incorporating textural features, lighting, damage, and augmentations to mimic real-world inscription diversity. Despite minimal real examples, we engineer a dataset of 250 000 training and 25 000 validation images covering the 22 letter classes in the Aramaic alphabet. This comprehensive corpus provides a robust volume of data for training a residual neural network (ResNet) to classify highly degraded Aramaic letters. The ResNet model demonstrates 95% accuracy in classifying real images from the 8th century BCE Hadad statue inscription. Additional experiments validate performance on varying materials and styles, proving effective generalization. Our results validate the model’s capabilities in handling diverse real-world scenarios, proving the viability of our synthetic data approach and avoiding the dependence on scarce training data that has constrained epigraphic analysis. Our innovative framework elevates interpretation accuracy on damaged inscriptions, thus enhancing knowledge extraction from these historical resources.

show abstract

Section: Machine Learning For Ancient Scriptsmentioning

confidence: 99%

Section: Training Datasets For Ancient Scriptsmentioning

confidence: 99%

Section: Synthetic Data For Ancient Scriptsmentioning

confidence: 99%

See 1 more Smart Citation

Deep Aramaic: Towards a synthetic data paradigm enabling machine learning in epigraphy

Aioanei,

Hunziker-Rodewald,

Klein

et al. 2024

PLoS ONE

View full text Add to dashboard Cite

show abstract

“…AI has made significant contributions to the field of CH, especially in deciphering ancient languages, restoring ancient text using deep learning, and automatic identification of objects. Furthermore, AI can assist in the restoration of damaged or incomplete ancient texts, using deep learning algorithms to fill in the missing parts (Rizk et al, 2021). In addition, AI can be used for the automatic identification of objects in CH collections, including point cloud and image classification and segmentation (Grilli et al, 2017).…”

Section: Ai For Chmentioning

confidence: 99%

Artificial Intelligence and Cultural Heritage: Design and Assessment of an Ethical Framework

Pansoni,

Tiribelli,

Paolanti

et al. 2023

Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

Abstract. The pioneering use of Artificial Intelligence (AI) in various fields and sectors, and the growing ethical debate about its application have led research centers, public and private institutions to establish ethical guidelines for a trustworthy implementation of these powerful algorithms. Despite the recognized definition of ethical principles for a responsible or trustworthy use of AI, there is a lack of a sector-specific perspective that highlights the ethical risks and opportunities for different areas of application, especially in the field of Cultural Heritage (CH). In fact, there is still a lack of formal frameworks that evaluate the algorithms’ adherence to the ethical standards set by the European Union for the use of AI in protecting CH and its inherent value. Because of this, it is necessary to investigate a different sectoral viewpoint to supplement the widely used horizontal approach. This paper represents a first attempt to design an ethical framework to embody AI in CH conservation practises to assess various risks arising from the use of AI in the field of CH. The contribution presents a synthesis of the different AI applications to improve the preservation process of CH. It explores and analyses in depth the ethical challenges and opportunities presented by the use of AI to improve CH preservation. In addition, the study aims to design an ethical framework of principles to assess the application of this ground-breaking technology at CH.

show abstract

A review of AI applications in human sciences research

Chapinal-Heras,

Díaz-Sánchez

2024

Digital Applications in Archaeology and Cultural Heritage

View full text Add to dashboard Cite

A Hybrid Capsule Network-based Deep Learning Framework for Deciphering Ancient Scripts with Scarce Annotations: A Case Study on Phoenician Epigraphy

Cited by 9 publications

References 6 publications

Deep Aramaic: Towards a synthetic data paradigm enabling machine learning in epigraphy

Deep Aramaic: Towards a synthetic data paradigm enabling machine learning in epigraphy

Artificial Intelligence and Cultural Heritage: Design and Assessment of an Ethical Framework

A review of AI applications in human sciences research

Contact Info

Product

Resources

About