Loading video...
Video Failed to Load
Before AI can understand the physical world, it needs data about the physical world. Not just images. Spatial environments. Real streets, buildings, parks, stores, museums, and public spaces captured in 3D. OVER has already reconstructed 290,000+ real-world 3D scenes, creating a growing dataset for the next generation of spatial... show more
24,045 views • 12 days ago •via X (Twitter)
23 Comments

Spatial data could be one of the biggest missing pieces for embodied AI. OVER is building a pretty interesting foundation here.

AI understanding the physical world needs more than pixels. Real 3D environments give models a much richer view of how the world actually works.

I've got thousands of hi-res photogrammetry scans (including raw images, drone, 360°), who wants to buy?

Use AI to convert street view images to 3D

AI understanding the real world starts with having real-world data.

Real streets and buildings give AI something much closer to the physical world.

OVR to the world 🌎

OVER is building serious spatial data infrastructure 🔥

Spatial data will matter more as AI moves into physical environments

290,000+ 3D scenes is a massive dataset. 👀

Spatial scenes add geometry and place context that image-only corpora miss.

Turning real environments into usable 3D data makes a lot of sense.

Spatial data could become very important for the next phase of AI.

The scale of those 3D scenes is impressive

I love your work, it’s outstanding 👏

This is a different side of AI that deserves more attention.

👏🙌

🎉

AI will need more than images to properly understand physical spaces.

Would Over be open to working with defense companies?

The robotics use case here is really interesting.

lol, no need to blur anybody - already auto blurred :D

Real world 3D data could become critical for embodied AI
