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
- Parking management challenges:
- Manual monitoring of parking slots
- No real-time occupancy tracking
- Inefficient space utilization
- Difficulty in:
- Detecting vehicle presence
- Identifying misaligned parking
Our Solution
- Developed a real-time parking detection system using computer vision
- Key capabilities:
- Vehicle detection using YOLO
- Parking slot mapping using coordinates
- Occupancy detection (free vs occupied)
- Alignment detection (proper vs misaligned)
Solution Architecture
- Frame capture module:
- Live camera or video input
- Detection module:
- YOLO-based vehicle detection
- Processing:
- Slot boundary comparison
- IoU and inside-percentage calculations
- Logic layer:
- Occupancy classification
- State stabilization
- Visualization:
- OpenCV overlays including boxes, masks, and warnings
Deliverables
- Real-time parking detection system
- Slot occupancy tracking
- Visualization output (OpenCV-based)
- Configurable environment setup
Technical Challenges
- Accurate detection in real-time video streams
- Defining parking boundaries precisely
- Handling misaligned vehicles
- Ensuring stable detection over time
Solutions
- Polygon-based slot mapping
- Time-based stabilization logic
- IoU and percentage-based calculations
- Dynamic overlays for visualization
Business Impact
- Improved parking space utilization
- Enabled real-time monitoring
- Reduced manual supervision
- Scalable for smart city applications
Tech Stack (Across All Projects)
Frameworks
- FastAPI
- Modal (serverless compute)
Languages / Techniques
- Python
- Computer vision
- ETL pipelines
- OCR processing
Models Used
- YOLOv8 (object detection)
- OCR models (Gemini, DeepSeek tested)
- Vision Language Models (experimental)
Skills Used
- Data processing and automation
- Computer vision
- System design
- Backend development
- Problem-solving
Databases
- JSON outputs and structured datasets
- No external database in most pipelines
Cloud / Infrastructure
- Modal serverless compute
- Local and containerized (Docker) setups
Overall Business Impact
- Automated previously manual engineering workflows
- Improved accuracy and reduced human error
- Enabled scalable processing of large datasets
- Delivered structured and actionable insights
- Reduced operational time and effort significantly
