App Link: https://structural-defect-detect.streamlit.app/
Structural Defect Detection is an AI-driven computer vision application designed to automatically identify, classify, and document construction defects from site images — improving safety, quality control, and operational efficiency across construction projects.
In large-scale construction projects, inspection of civil structures such as: Buildings, Bridges and Industrial facilities is typically performed through manual visual checks.
This traditional approach suffers from:
- High dependency on human expertise and judgment
- Subjective and inconsistent defect identification
- Delayed discovery of structural issues
- Poor standardization of reporting formats
- Increased safety risks and rework costs
Common defects such as cracks, honeycombing, uneven plaster, and surface deformities often go unnoticed until advanced stages of construction.
To develop an AI-powered vision-based system that automates:
- Structural defect detection
- Defect classification and severity assessment
- Standardized documentation and reporting
The system aims to provide:
- Accurate and consistent inspection results
- Faster turnaround during active construction phases
- Scalable integration into quality assurance workflows
- Decision support for site engineers and project managers
An end-to-end AI-powered structural inspection platform built using:
- Convolutional Neural Networks (CNNs) for defect detection
- Google Gemini for multimodal visual understanding
- Streamlit for intuitive user interaction
- Upload construction site images
- AI analyzes visual defects
- Defects are classified and evaluated
- Structured reports are generated automatically
- 📷 Image upload from construction sites
- 🔍 Automated detection of:
- Concrete cracks
- Honeycombing
- Plaster inconsistencies
- Surface irregularities
- 📊 Structured defect classification:
- Defect type
- Location (floor / zone)
- Severity (Low / Medium / High)
- 🧾 Automated corrective action suggestions
- 📄 Downloadable inspection reports (PDF / Word)
- ⚡ Real-time inference with cloud deployment
| Component | Technology |
|---|---|
| Frontend | Streamlit |
| Backend | Python |
| AI / Vision | CNN (Custom Pretrained Models) |
| LLM Integration | Google Gemini API |
| Deployment | Streamlit Cloud |
| Document Output | python-docx, xhtml2pdf |
| Version Control | Git & GitHub |
- 75% reduction in time spent on manual quality inspections
- Early defect identification reduces costly rework
- Improved consistency and objectivity across multiple sites
- 80% reduction in manual report-writing effort
- Scalable quality control across projects
- EPC contractors
- Civil engineering consultants
- Real estate developers
- Infrastructure inspection teams
- Quality assurance departments