AI-Powered 2D Floor Plan to 3D Visualization Platform

Intelligent 2D Plan Parser

Built a computer vision pipeline that accepts JPG, PNG, and PDF uploads of floor plans across wildly varied styles, including hand-drawn sketches, scanned paper blueprints, and digital CAD exports. The parser handles noise, skew correction, and varying line weights to extract a clean semantic representation of the space.

AI-Driven Room & Wall Segmentation

Deployed a CV segmentation model trained on architectural floor plan datasets to detect walls, rooms, doors, and windows with high precision. It also supports curved walls, open-plan spaces, and multi-storey layouts. Each detected element is classified and assigned geometry for 3D reconstruction.

Automated 3D Geometry Reconstruction

From the segmented floor plan, the pipeline extrudes walls to standard ceiling heights, places door and window apertures, and generates room geometry. Users can adjust ceiling heights, apply material presets, and configure lighting in the browser.

Real-Time Browser-Based 3D Viewer & Export

The output is rendered in a real-time Three.js and Babylon.js 3D viewer, with no plugin and no download. Users can orbit, zoom, and walk through the space. One-click export to PNG, GLB, and PDF supports listings and client presentation workflows.

Architecture Firms

Instantly convert any client-submitted floor plan sketch into a walkable 3D model for design review meetings, no BIM specialist required

Interior Design Studios

Visualize proposed room layouts, material choices, and furniture placement in 3D from a hand-drawn concept sketch in seconds

Real Estate Developers

Generate interactive 3D walkthroughs of off-plan properties from architect-supplied 2D drawings for pre-sales listings and investor decks

Construction Project Managers

Produce rapid 3D site visualizations from scanned legacy blueprints to communicate spatial changes to non-technical stakeholders