Deep learning-driven digital twin modeling and immersive intelligent virtual exhibition for cultural relics

Cultural heritage preservation increasingly uses digital systems for accurate representation of artifacts. These systems enable interactive access through immersive virtual environments. Traditional methods rely on manual or single-modal approaches, limiting scalability and integration. This research proposes a deep learning (DL)-based model for high-fidelity digital twin modeling. It integrates bio-inspired optimization with transformer-based multimodal fusion to enhance performance and representation. A sterna migration optimized multimodal fusion transformer network (SM-MFT Net) is designed to improve representation learning and decision-making. A multimodal dataset comprising artifact images ( n = 5768), multi-view three-dimensional (3D) representations ( n = 5768), and associated metadata ( n = 5768) is utilized. After data collection, the cultural relics dataset undergoes preprocessing, where image data is enhanced using Gaussian filtering (GF) and textual data is processed through tokenization to improve data quality. Subsequently, EfficientNet is employed to extract discriminative visual features, while principal component analysis (PCA) is used to derive compact semantic representations for effective multimodal learning. The extracted features are fused through an MFT Net, while the SM Algorithm optimizes model parameters and fusion weights for improved performance. Experimental results show that the proposed SM-MFT Net achieves CD = 0.0150, F-Score = 0.7000, NC = 0.9200, PSNR = 17.00 dB, SSIM = 0.8200, DR = 0.750, and VQ = 0.450, demonstrating its effectiveness for high-fidelity digital twin generation and immersive virtual heritage exhibitions. The proposed model demonstrates strong capability in integrating multimodal data for intelligent digital twin generation and immersive cultural heritage exhibition.

Authors

Institutions

Publication Details

Journal
Discover Artificial Intelligence
Published
2026-11-21
DOI
https://doi.org/10.1007/s44163-026-02124-z
Primary Topic
3D Surveying and Cultural Heritage
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Deep learning-driven digital twin modeling and immersive intelligent virtual exhibition for cultural relics

Tong Xu
Discover Artificial Intelligence
3D Surveying and Cultural Heritage
article

Deep learning-driven digital twin modeling and immersive intelligent virtual exhibition for cultural relics

Tong Xu
article en

Abstract

Cultural heritage preservation increasingly uses digital systems for accurate representation of artifacts. These systems enable interactive access through immersive virtual environments. Traditional methods rely on manual or single-modal approaches, limiting scalability and integration. This research proposes a deep learning (DL)-based model for high-fidelity digital twin modeling. It integrates bio-inspired optimization with transformer-based multimodal fusion to enhance performance and representation. A sterna migration optimized multimodal fusion transformer network (SM-MFT Net) is designed to improve representation learning and decision-making. A multimodal dataset comprising artifact images ( n = 5768), multi-view three-dimensional (3D) representations ( n = 5768), and associated metadata ( n = 5768) is utilized. After data collection, the cultural relics dataset undergoes preprocessing, where image data is enhanced using Gaussian filtering (GF) and textual data is processed through tokenization to improve data quality. Subsequently, EfficientNet is employed to extract discriminative visual features, while principal component analysis (PCA) is used to derive compact semantic representations for effective multimodal learning. The extracted features are fused through an MFT Net, while the SM Algorithm optimizes model parameters and fusion weights for improved performance. Experimental results show that the proposed SM-MFT Net achieves CD = 0.0150, F-Score = 0.7000, NC = 0.9200, PSNR = 17.00 dB, SSIM = 0.8200, DR = 0.750, and VQ = 0.450, demonstrating its effectiveness for high-fidelity digital twin generation and immersive virtual heritage exhibitions. The proposed model demonstrates strong capability in integrating multimodal data for intelligent digital twin generation and immersive cultural heritage exhibition.

Discover Artificial IntelligenceVol. 6(1)
Shaoxing University (CN)
3D Surveying and Cultural Heritage
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon