Automated floor plan auditing platform
Emad عماد
Space, understood.
From blueprint to structured audit. Emad combines YOLO-based spatial detection with Generative AI to check architectural floor plans against Saudi Building Code requirements.
- Input
- PDFs, CAD exports, and high-resolution blueprint images.
- Output
- Structured audit logs with compliance status and identified code violations.
Explore the architecture: Emad
From drawing to decision
- Receive the blueprint
Asynchronous FastAPI endpoints accept blueprint uploads and orchestrate the computer vision and Generative AI pipeline.
- Detect and measure
YOLO detects and segments walls, doors, windows, corridors, and egress routes. Bounding boxes isolate zones for spatial calculations.
- Cross-reference the code
Generative AI compares extracted geometry—including room dimensions, corridor widths, and exit distances—with regulatory text from SBC guidelines.
- Return the findings
The API returns structured JSON audit logs containing compliance status and specific code violations.
The engineering challenge
Connecting visual elements in a drawing to the measurements and regulatory language needed for a useful audit.
Architecture
A hybrid CV/GenAI pipeline coordinated through asynchronous REST endpoints. FastAPI handles file intake, pipeline orchestration, and structured responses.
The deliverable
An automated auditing platform that turns uploaded architectural drawings into inspectable compliance findings.