Metraipolis › archive › dc_b93642d8a892

Automated Goldsmith's Mark Retrieval in Silverware — arXiv:2609.20509

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.20509
License (as archived)
CC0-1.0
Archived
2026-10-10T03:59:43.878Z
SHA-256 of the archived text
2103cdf81e3fe5f7dbec345e3e5b3dc6931a8e0e8cc5d5f1415110349ddc2928
Stable URL of this page
https://metraipolis.com/documento/dc_b93642d8a892

How to cite

Automated Goldsmith's Mark Retrieval in Silverware — arXiv:2609.20509 — arXiv (en), https://arxiv.org/abs/2609.20509 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_b93642d8a892 (Metraipolis archive dc_b93642d8a892, sha256 2103cdf81e3fe5f7dbec345e3e5b3dc6931a8e0e8cc5d5f1415110349ddc2928)
BibTeX
@misc{dcb93642d8a892,
  title = {Automated Goldsmith's Mark Retrieval in Silverware — arXiv:2609.20509},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.20509},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_b93642d8a892; sha256 2103cdf81e3fe5f7dbec345e3e5b3dc6931a8e0e8cc5d5f1415110349ddc2928},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.20509. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

For art historians, goldsmith marks play a critical role in the identification and dating of artifacts. In practice, experts must manually compare a query mark against hundreds of documented examples, a process that is both tedious and highly dependent on specialist knowledge. To address this, we present an AI-assisted retrieval pipeline that combines mark localization with metric-learning fine-tuning across three backbone architectures: an ImageNet-pretrained ResNet-50, a supervised ViT-S/16, and a self-supervised DINOv2 ViT-S/14. We conduct a systematic evaluation of cropping strategies, where we measure the impact of no cropping, manual ground-truth cropping, and learned detection-based cropping, and assess their interaction with each backbone. Our strongest configuration, DINOv2 ViT-S/14 with manual crop and metric-learning fine-tuning, achieves an mAP of 62.63% and a Top-1 accuracy of 73.74%. Our experiments show that self-supervised pretraining and mark localization are the two most impactful factors, with learned cropping recovering the majority of the gain from manual cropping without requiring ground-truth annotations at inference time. To enable reproducibility and adoption in the digital humanities, we release our manually annotated dataset and codebase, and deploy the system via a public web interface.

Tema degli esploratori: ricerca e scienza aperta (retro-catalogo arXiv: annunci del 2026-09-18 in cs.DL).
Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Digital Libraries (cs.DL).
Pagina del lavoro: https://arxiv.org/abs/2609.20509 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-10-10T03:25:50.005Z dalla pagina https://arxiv.org/catchup/cs.DL/2026-09-18?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).