spatial-ai

🐍 Geometric Computer Vision Library for Spatial AI

InternRobotics' open platform for building generalized navigation foundation models.

Official implement of VGGT-Long

Generalizable Perception Stack for all things 3D, 4D & Scene Understanding

Terra: Hierarchical Terrain-Aware 3D Scene Graph for Task-Agnostic Outdoor Mapping

Real-Time Spatio-Semantic Memory

MLX-native 3D and spatial inference for Apple silicon: object reconstruction, image-to-mesh, scene geometry, Gaussian splats, and multi-view bundles.

A Synthetic Benchmark for Evaluating Spatial Intelligence in Visual Foundation Models

Spatial analysis using Deep Spatio Temporal Point Process (DeepSTPP) updated with state-of-the-art mamba state space model, using spatiotemporal Hawkes Process, spatiotemporal self-correcting process, NJ COVID-19 and Japanese earthquake data.

Feed-forward video-to-3D scene reconstruction using VGGT (CVPR 2025) with open-vocabulary semantic labeling via SAM 2.1 + CLIP. One command, no COLMAP, Apple Silicon + CUDA.

3D scene reconstruction from monocular video — COLMAP + 3D Gaussian Splatting + Grounded-SAM-2, with open-vocabulary CLIP queries and an agentic QA release gate.

Stochastic Preconditiong for Neural Field Optimization -- additional website assets

A curated benchmark zoo for embodied navigation tasks, datasets, metrics, leaderboards, and reproducibility notes.

repo to describe the floor layout project

Source-backed guide to AI world models, generated worlds, spatial AI, and model release signals

Real-time 3D Hand & Body Pose Estimation with NVIDIA FoundationStereo & RTMPose for Dual-Eye Intelligence Perception (DIP)

Autonomous agent for SF design-build firms: finds AI-homeowner leads, generates 3D remodel proposals, drafts outreach, and learns from outcomes.

AI-powered indoor spatial navigation — Where Google Maps ends, Inevi begins. Live camera location detection, conversational AI guide, multilingual (EN/TE/HI), built with Groq, AWS Aurora DSQL, DynamoDB, S3, and Vercel.

Reproduce Neural Assets in PyTorch with a clean, paper-faithful implementation for MoVi training and research

Augmenta — independent third-party profile of a public API surface, by API Evangelist. Augmenta is a Toronto-based generative AI company building the Augmenta Construction Platform (ACP), a Spatial AI system that automates building design — starting with electrical raceway routing and expanding into mechanical and plumbing (MEP).