Click2Poly VLM Speeds Up Geospatial Vector Mapping

Nicolas Girard, Jawher Ben Abdallah, Arno Gobbin, Liuyun Duan, Sacha Lepretre· August 13, 2026 View original

Key takeaways

  • Manual vector mapping is a bottleneck in geospatial data production.
  • Click2Poly is an AI assistant that uses a VLM to speed up manual editing.
  • It allows direct vector layer editing via user clicks within QGIS.
  • The tool significantly improves efficiency for cartographers in production environments.

Who benefits

GeospatialUrban PlanningConstructionUtilitiesGovernment

Summary

Click2Poly is a human-in-the-loop AI assistant that extends the Florence-2 Vision Language Model (VLM) to accelerate the manual editing of building and wall vector layers. Implemented as a QGIS plugin, it allows users to edit geospatial data directly with clicks, significantly improving efficiency in production environments.

Accurate vector mapping of buildings and walls is a crucial but labor-intensive task in geospatial applications. While deep learning has advanced automatic extraction, human quality control and manual correction of complex cases remain indispensable to meet cartographic standards. This paper introduces Click2Poly, an AI assistant designed to streamline this manual editing process. Click2Poly operates as a human-in-the-loop system, building upon the Florence-2 Vision Language Model (VLM). It allows users to interact with the system by simply clicking, and in response, the AI directly edits the building or wall vector layer. This intuitive interaction aims to significantly reduce the time and effort required for corrections. The tool is implemented as a plugin for QGIS, a widely used open-source geographic information system. Its integration into a real-world production environment demonstrates its capability to speed up the manual editing of geospatial vector data, making the overall mapping workflow more efficient and less burdensome for cartographers and GIS professionals.

Why it matters

Professionals in GIS, urban planning, and infrastructure development can leverage Click2Poly to drastically reduce the time and cost associated with creating and maintaining accurate geospatial vector data, improving project timelines and resource allocation.

How to implement this in your domain

  1. 1Evaluate Click2Poly as a QGIS plugin for current geospatial mapping projects involving building and wall vectorization.
  2. 2Train GIS specialists and cartographers on how to effectively use the human-in-the-loop AI assistant for quality control and editing.
  3. 3Integrate the tool into existing geospatial data production workflows to identify bottlenecks and measure efficiency gains.
  4. 4Provide feedback to the developers for potential enhancements or customizations specific to organizational needs.

Original post by Nicolas Girard, Jawher Ben Abdallah, Arno Gobbin, Liuyun Duan, Sacha Lepretre

"arXiv:2608.11424v1 Announce Type: new Abstract: Accurate vector mapping of buildings and walls is critical for geospatial applications but remains a labor-intensive process. While recent deep learning methods have improved automatic extraction, in order to meet cartographic stand…"

View on X

Originally posted by Nicolas Girard, Jawher Ben Abdallah, Arno Gobbin, Liuyun Duan, Sacha Lepretre on X · view source

Want to go deeper?

Turn these trends into skills with Learnijoy's hands-on AI & tech courses.

Explore courses