← Back to all news
arXiv Computer Vision

Generalizable Multiscale Segmentation of Heterogeneous Map Collections

The paper introduces Semap, an open benchmark dataset of 1,439 manually annotated map patches that captures the diversity of historical map collections. It proposes a generalizable semantic segmentation workflow that combines procedural data synthesis with multiscale integration, achieving the best performance among evaluated methods on heterogeneous map data. The study demonstrates that segmentation accuracy remains stable across different map styles, scales, geographic regions, and publication contexts, thereby facilitating the integration of diverse cartographic archives into historical geographic research.

Related stories

arXiv AI
Oct 27

ArchiveLens: A Generative Spatial Knowledge Framework for Semantic Reconstruction of Historical Maps

ArchiveLens is a generative spatial knowledge framework that transforms historical maps into entity‑centric, machine‑queryable geospatial databases. The pipeline separates morphological and semantic reconstruction into three stages: building footprint extraction with tiled YOLOv11‑seg, deterministic material classification using HSV color thresholds, and localized attribute parsing via a quantized vision‑language model with domain‑specific prompts. Applied to 445 Sanborn map sheets of Chicago (1918‑1950), the system delivers high geometric and semantic accuracy and enables researchers to query building attributes and compare spatial patterns across time.

By Yanan Wu, Yalin Yang, Chengbin Deng
llmsragcomputer-visionefficiencymultimodal
More like this →
arXiv Computer Vision
1d ago

TerraMesh-Masks: Open-Vocabulary Segmentation for Earth Observation

TerraMesh-Masks is a newly released large-scale dataset for open-vocabulary segmentation in Earth observation, containing 16 million binary masks aligned with nine million TerraMesh samples and annotated with OpenStreetMap and Overture semantics. The dataset includes an expert-reviewed evaluation set of 523 samples covering over 200 classes. Using TerraMesh-Masks, the authors built an open-vocabulary version of TerraMind that outperforms general-purpose models, achieving 35.18% IoU compared to 16.29% for DINOv3 and 11.33% for SAM-3.

By Benedikt Blumenstiel, Hugues Devimeux, Johannes Jakubik, Konrad Schindler
computer-visionfine-tuningmultimodalbenchmarks
More like this →
arXiv AI
Sep 1

AtlasNLP: A Country-Aware Atlas of Dataset Representation in NLP

arXiv:2608.30107v1 Announce Type: cross Abstract: Understanding which countries are represented in NLP datasets is essential for identifying gaps, targeting data collection, measuring progress, and i...

By Joan Nwatu, Tsedeniya Solomon Amare, Longju Bai, Bontu Fufa Balcha, Zayd Bashir, Angana Borah, Zara Burzo, Yubin Choi, Naihao Deng, Samika Gupta, Michel Faloughi, Claude Kwizera, Ziqiao Ma, Cynthia Yacel Fuertes Panizo, Ellie Seehorn, Hui Shen, Jiayi Tang, Zesen Zhao, Boyuan Zheng, Rada Mihalcea
nlp
More like this →
arXiv Computer Vision
Sep 18

Automated Goldsmith's Mark Retrieval in Silverware

The paper introduces an AI-assisted pipeline for retrieving goldsmith marks from silverware images, combining mark localization with metric‑learning fine‑tuning on three backbone models (ResNet‑50, ViT‑S/16, and DINOv2 ViT‑S/14). Systematic tests of cropping strategies show that manual cropping and metric‑learning fine‑tuning yield the best performance, with DINOv2 ViT‑S/14 achieving 62.63% mAP and 73.74% Top‑1 accuracy. The authors release a manually annotated dataset, code, and a public web interface to support reproducibility and adoption in digital humanities.

By Atmik Tiwari, Vincent Christlein, Mark Fichtner, Freya Gohlke, Birgit Sch\"ubel, Theresa Witting, Heike Zech, Mathias Zinnen
fine-tuning
More like this →