Blog Posts

From 3D Semantic Search to 3D Object Detection

A natural language query becomes a 3D object. We extend our semantic search pipeline with prompt-based segmentation and multi-view visual hull computation to automatically recover the geometry of detected objects. Continue reading →

Searching the Physical World: Bridging 3D Models and LMMs

High-fidelity 3D reconstruction solves the problem of visualization, but not of understanding. Integrating Large Multimodal Models (LMMs) allows us to connect natural language to spatial geometry, enabling semantic search and localization within 3D environments. Continue reading →