Client Background
- Client Name: A leading AI Data Center firm in the USA
- Industry Type: Engineering Analytics & AI Solutions
- Products & Services:
- Engineering document analysis
- Computer vision-based automation
- AI-driven validation systems
About Client:
- The client works with complex engineering datasets including:
- Electrical panel images
- Architectural drawings
- Multi-page technical PDFs
- Their workflows required high accuracy, scalability, and automation
- Objective:
- Reduce manual inspection
- Improve validation accuracy
- Automate structured data extraction
The Problem
- Engineering documents contained hundreds of sheets in multi-page PDFs
- Manual verification issues:
- Sheet index mismatch with actual pages
- Inconsistent naming formats
- Difficult to validate document structure
- High time consumption and risk of errors
Our Solution
- Built an automated PDF validation system
- Key features:
- Extract sheet indices using regex
- Detect document type automatically
- Validate sheet order and structure
- Generate structured JSON reports
- Integrated:
- OCR and Vision Language Models
- YOLO for symbol detection
Solution Architecture
- User uploads PDF via FastAPI
- Modal compute handles:
- PDF ingestion
- Page extraction
- Image tiling (3×3 grid)
- YOLOv8:
- Symbol detection
- OCR (Gemini):
- Text extraction
- Validation engine:
- Regex-based sheet extraction
- Cross-verification logic
- Output:
- JSON report
- Match percentage
- Validation results
Deliverables
- Automated PDF analysis system
- Sheet index validation engine
- JSON-based structured reporting
- Symbol detection model
Technical Challenges
- Extracting structured data from unstructured PDFs
- Handling variations in document formats
- Spatial understanding of engineering drawings
- OCR accuracy on complex layouts
Solutions
- Custom regex patterns for sheet extraction
- Hybrid OCR and VLM approach
- YOLO-based symbol detection
- Batch processing for scalability
Business Impact
- Eliminated manual document validation
- Improved accuracy of sheet verification
- Reduced processing time significantly
- Enabled scalable document analysis
