text-on-maps-viz

This repository contains code and scripts to visualize text detected on maps from the David Rumsey Map Collection. To get there, we take the following steps:

1. Convert georeference data from the David Rumsey Map Collection into Georeference Annotations.
2. Convert the text detected with OCR produced by mapKurator into OCR Web Annotations and add the IIIF Image ID from the Georeference Annotations to this data.
3. Use Allmaps to turn the pixel coordinates of the text bounding boxes into GeoJSON.
4. Turn this GeoJSON into PMTiles using tippecanoe.
5. Visualize this data in a web application built with SvelteKit and MapLibre GL JS .

This repository contains the following directories:

• etl: ETL scripts to transform and export the required data.
• 'app`: Web application to visualize the data.
• data: Input and output data.

Getting Started

To run the scripts or app locally, first install the required dependencies:

pnpm install --recursive

Then, run the ETL scripts to produce the required data (or download them from Zenodo) and build the web application.

Prerequisites

Software:

• Node.js v23.1.0 or higher
• pnpm v10.10.0 or higher
• Tippecanoe

Required input data:

• ./data/input/rumsey_57k_english.zip (warning: This is a 51 GB file!)
• ./data/input/maps.ndjson, produced by https://github.com/allmaps/rumsey-scripts.

See the etl directory for more details.