OVER Expands Global VPS Coverage with Mapillary and Panoramax Integration

We’re thrilled to announce that OVER is expanding its Visual Positioning System (VPS) coverage globally, incorporating areas from the open datasets of Mapillary and Panoramax. Precise re-localization—everywhere!

Mapillary and Panoramax are two prominent crowdsourced street-level imagery platforms, together providing access to an immense combined database of over 2 billion geolocalized pictures available under the CC-BY-SA license. Mapillary, recently acquired by Meta, offers extensive global coverage, while Panoramax primarily focuses on France. To get a sense of their impressive scale, you can check out Mapillary’s global coverage map here:

Their imagery data complements the extensive datasets collected by OVER’s map2earn program. While Mapillary and Panoramax provide widespread global coverage, their data density is significantly sparser compared to OVRMaps (approximately 1-2 orders of magnitude sparser). Additionally, the quantity and quality of imagery vary widely across different locations.

These dataset characteristics present formidable technical challenges for building a robust VPS service, including limited overlapping points of view, heterogeneous camera distortions, and the absence of metric scaling. Despite these hurdles, recent months have brought significant breakthroughs. By integrating our VPS pipeline with cutting-edge Large Geospatial Models (LGMs), we’ve successfully developed methods to create accurate 3D digital twins from this data and deliver reliable VPS re-localization services for these expansive yet sparse datasets. Although accuracy levels on these new datasets will be somewhat lower compared to our community-driven OVRMaps, they still vastly surpass the accuracy achievable with standard GPS.

This advancement represents a significant leap forward for OVER’s platform utility. It allows us to dramatically extend our VPS coverage beyond the impressive 110,000+ locations already mapped by our community, vastly surpassing the VPS coverage offered by our largest Web2 competitor, Niantic. This highlights how decentralized physical infrastructure networks (DePIN) combined with open-source data can effectively outperform traditional Web2 models.

But that’s not all. The integration of these open-source, sparse datasets also unlocks two exciting new opportunities:

1. Expanding Map2Earn Data Acquisition Techniques: The newly developed data processing pipeline will enable faster and more efficient mapping on OVER through the integration of 360-degree camera mapping.
2. Integrating New Data Providers: Imagery datasets captured by different DePIN operators can now seamlessly integrate into the OVER VPS pipeline, exemplified by our recent partnership with Natix.

Stay tuned for more exciting updates!