Cadastral-Grounded Geospatial Dataset for Fine-Grained Spatial Understanding

GroundSet is a large-scale Earth Observation dataset built on 20 cm resolution optical aerial orthophotos and legally verified cadastral vector data from the French national mapping agency (IGN). It is designed to advance fine-grained spatial understanding for multimodal models. The dataset was created by RogerFerrod and was last updated in March 2026.

Use Cases

• Train object detection models to identify property boundaries and land parcels using the cadastral vector data overlayed on high-resolution orthophotos.
• Fine-tune visual question answering models for spatial reasoning tasks based on the semantic richness of the combined image and vector data.
• Develop zero-shot image classification systems for land use and land cover categories using the verifiable ground truth from the cadastral data.
• Benchmark image segmentation models for precise delineation of cadastral features against the legally verified vector data.

Strengths

• Dataset is built on high-resolution 20 cm optical aerial orthophotos.
• Vector data is legally verified and provided by the French national mapping agency (IGN).
• Dataset is categorized as large-scale, with a size category of 100K-1M samples.

Limitations

• Specific row count, column details, and file formats are unknown.
• Geographic coverage is limited to the region of France, as indicated by the source agency.
• The dataset's license and exact update frequency are not specified in the provided input.

Provenance

Source

French national mapping agency (IGN).

Collection Method

Built upon high-resolution optical aerial orthophotos and legally verified cadastral vector data.

Freshness

Last updated on 2026-03-17.

Geography

France (implied by source agency IGN).