Automated Archaeological Image Annotation: AI-Assisted Object Recognition and Metadata Enrichment

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F67985912%3A_____%2F25%3A00638695" target="_blank" >RIV/67985912:_____/25:00638695 - isvavai.cz</a>

Nalezeny alternativní kódy

RIV/68081758:_____/25:00638695

Výsledek na webu

<a href="https://drive.google.com/file/d/1Q9IG_LuJRkYHwwHawe7IpIEqAqT7Vtuh/view" target="_blank" >https://drive.google.com/file/d/1Q9IG_LuJRkYHwwHawe7IpIEqAqT7Vtuh/view</a>

DOI - Digital Object Identifier

—

Jazyk výsledku

angličtina

Název v původním jazyce

Automated Archaeological Image Annotation: AI-Assisted Object Recognition and Metadata Enrichment

Popis výsledku v původním jazyce

The ongoing digitisation of archaeological image archives presents significant opportunities for knowledge discovery, yet also poses considerable challenges, as processing vast amounts of visual data remains a time-intensive task well suited for automation. The application of artificial intelligence (AI) and distant viewing methods offers a scalable solution to enhance the usability, accessibility, and interoperability of large archaeological image archives. Without such automation, achieving comparable results would require years of manual processing. This paper presents a workflow for automatic image annotation, developed to improve (meta)data quality in the Archaeological Map of the Czech Republic (AMCR) repository and discovery services. We outline the training process and pilot implementation of a deep learning model fine-tuned for archaeological datasets, employing ResNet architecture. The workflow enables segmentation and annotation of archaeological images using domain-specific controlled vocabulary terms, facilitating the identification of artefact types and other relevant visual elements. To address the diversity of archaeological photography, we train the model on two distinct image categories: (1) single artefact/find images, typically photographed on standardised backgrounds with scales, and (2) excavation and fieldwork photographs, capturing a wide range of archaeological contexts, from entire excavations and sites to individual trenches and burials.

Název v anglickém jazyce

Automated Archaeological Image Annotation: AI-Assisted Object Recognition and Metadata Enrichment

Popis výsledku anglicky

The ongoing digitisation of archaeological image archives presents significant opportunities for knowledge discovery, yet also poses considerable challenges, as processing vast amounts of visual data remains a time-intensive task well suited for automation. The application of artificial intelligence (AI) and distant viewing methods offers a scalable solution to enhance the usability, accessibility, and interoperability of large archaeological image archives. Without such automation, achieving comparable results would require years of manual processing. This paper presents a workflow for automatic image annotation, developed to improve (meta)data quality in the Archaeological Map of the Czech Republic (AMCR) repository and discovery services. We outline the training process and pilot implementation of a deep learning model fine-tuned for archaeological datasets, employing ResNet architecture. The workflow enables segmentation and annotation of archaeological images using domain-specific controlled vocabulary terms, facilitating the identification of artefact types and other relevant visual elements. To address the diversity of archaeological photography, we train the model on two distinct image categories: (1) single artefact/find images, typically photographed on standardised backgrounds with scales, and (2) excavation and fieldwork photographs, capturing a wide range of archaeological contexts, from entire excavations and sites to individual trenches and burials.

Druh

O - Ostatní výsledky

CEP obor

—

OECD FORD obor

60102 - Archaeology

Projekt

—

Návaznosti

I - Institucionalni podpora na dlouhodoby koncepcni rozvoj vyzkumne organizace

Rok uplatnění

2025

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů