Moscow Geolocation Index with 8,624 Panoramas

Moscow-1Km provides a pre-computed MegaLoc feature descriptor index for visual geolocation within a 1.0 km radius of central Moscow. The dataset contains 8,624 panorama images, resulting in 34,496 index entries, each with a 1024-dimensional descriptor. It was created by netryx-hub for use with the Netryx Drishti platform.

Use Cases

• Perform visual place recognition by matching query images against the 34,496 MegaLoc descriptor entries.
• Benchmark geolocation algorithms using the 8,624 panorama images centered on Moscow coordinates.
• Analyze the distribution of 1024-dimensional feature descriptors across the 1.0 km coverage area.

Strengths

• Contains 8,624 panorama images, providing substantial visual coverage.
• Offers 34,496 indexed feature entries for geolocation queries.
• Utilizes high-dimensional MegaLoc descriptors reduced to 1024 dimensions via PCA.

Limitations

• Limited geographic scope, covering only a 1.0 km radius in Moscow.
• Dataset utility is tightly coupled with the proprietary Netryx Drishti platform.
• No raw image data or standard tabular columns are provided, only pre-computed indices.

Provenance

Source

netryx-hub on Hugging Face.

Collection Method

Pre-computed index from panorama images using the MegaLoc descriptor model.

Freshness

Last updated on 2026-03-22.

Geography

Covers a 1.0 km radius centered at coordinates 55.756900, 37.615100 in Moscow, Russia.