Self-supervised Pre-training for Mapping of Archaeological Stone Wall in Historic Landscapes Using High-Resolution DEM Derivatives

2025-10-20

AI Summary

This study addresses the challenge of automated dry-stone wall detection in archaeological landscapes under conditions of dense vegetation occlusion and scarce annotated data. We propose a self-supervised semantic segmentation method leveraging multi-view high-resolution LiDAR-derived digital elevation models (DEMs). Our key innovation is the DINO-CV cross-view pretraining framework, which employs knowledge distillation to learn geometrically invariant features across diverse DEM derivatives—including slope, curvature, and topographic shadow—marking the first application of cross-view self-supervised learning to cultural heritage remote sensing mapping. The method is compatible with mainstream backbones (e.g., ResNet, ViT) and achieves 68.6% mean Intersection-over-Union (mIoU) on the Budj Bim World Heritage site. With only 10% labeled data for fine-tuning, it attains 63.8% mIoU, substantially reducing reliance on manual annotations. Results demonstrate strong efficacy and generalizability for low-resource historical landscape analysis.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Multi-instance/Multi-view Learning
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Web Mining and Content Analysis: Large pretrained models with web data
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Dry-stone walls hold significant heritage and environmental value. Mapping these structures is essential for ecosystem preservation and wildfire management in Australia. Yet, many walls remain unidentified due to their inaccessibility and the high cost of manual mapping. Deep learning-based segmentation offers a scalable solution, but two major challenges persist: (1) visual occlusion of low-lying walls by dense vegetation, and (2) limited labeled data for supervised training. We propose DINO-CV, a segmentation framework for automatic mapping of low-lying dry-stone walls using high-resolution Airborne LiDAR-derived digital elevation models (DEMs). DEMs overcome visual occlusion by capturing terrain structures hidden beneath vegetation, enabling analysis of structural rather than spectral cues. DINO-CV introduces a self-supervised cross-view pre-training strategy based on knowledge distillation to mitigate data scarcity. It learns invariant visual and geometric representations across multiple DEM derivatives, supporting various vision backbones including ResNet, Wide ResNet, and Vision Transformers. Applied to the UNESCO World Heritage cultural landscape of Budj Bim, Victoria, the method identifies one of Australia's densest collections of colonial dry-stone walls beyond Indigenous heritage contexts. DINO-CV achieves a mean Intersection over Union (mIoU) of 68.6% on test areas and maintains 63.8% mIoU when fine-tuned with only 10% labeled data. These results demonstrate the potential of self-supervised learning on high-resolution DEM derivatives for automated dry-stone wall mapping in vegetated and heritage-rich environments with scarce annotations.

Problem

• Mapping dry-stone walls in vegetated landscapes using DEMs
• Overcoming visual occlusion from dense vegetation with LiDAR data
• Addressing limited labeled data via self-supervised pre-training

Innovation

• Self-supervised pre-training for archaeological wall mapping
• Using high-resolution DEM derivatives to overcome vegetation occlusion
• Cross-view knowledge distillation for limited labeled data