Overview

Survai is a Smart Site Scanner Pipeline that transforms 3D scans from KIRI Engine (iOS LiDAR/photogrammetry) into intelligent annotated models for Unreal Engine 5.4+ PCG integration and MEP (Mechanical, Electrical, Plumbing) planning.

Architecture

  • Backend: Python 3.10+, FastAPI
  • Database: PostgreSQL
  • ML Models: YOLOv8, SAM, PointNet++
  • Frontend: React 18 dashboard
  • UE5 Plugin: PCG integration for spline routes
  • Deployment: Docker + Azure DevOps CI/CD

Pipeline

  1. KIRI Engine 3D scan import (iOS LiDAR/Photogrammetry)
  2. UV texture extraction
  3. 2D detection (YOLOv8/SAM) on UV maps
  4. 3D detection (PointNet++) on point clouds
  5. Detection fusion (2D + 3D)
  6. Utility route estimation (electrical, plumbing, HVAC)
  7. MEP catalog selection (19 catalogs, NEC/IPC standards)
  8. Export (JSON/CSV/GeoJSON/Splines for UE5)

Key Features

  • Auto-material selection and cost calculation
  • Mesh Analyzer with AI-assisted classification
  • 19 MEP catalogs with NEC/IPC compliance