>
Intelligent Insurance Evidence Review using Computer Vision & Google Gemini AI
Analyze damage images, verify claim evidence, assess fraud risk, generate explainable AI insights, and create professional claim reports—all from a single platform.
https://claimvision-ai-vq2afjukbbvckxmkbuhdil.streamlit.app/
⭐ If you like this project, don't forget to Star the repository!
ClaimVision AI is an AI-powered damage claim verification platform designed to streamline insurance claim review through Computer Vision and Large Language Models.
Instead of manually inspecting every claim, reviewers can upload damage evidence and receive AI-assisted insights, including:
- Damage detection
- Image quality validation
- Evidence consistency analysis
- Fraud risk assessment
- Explainable AI reasoning
- Professional PDF reports
Unlike systems limited to vehicle insurance, ClaimVision AI is built as a general-purpose damage verification platform capable of analyzing multiple categories of insurance claims.
| Category | Examples |
|---|---|
| 📦 Package Damage | Torn boxes, crushed parcels, damaged packaging |
| 💻 Electronics | Laptop, monitor, TV, tablet, keyboard |
| 📱 Mobile Devices | Smartphones, accessories |
| 🚗 Vehicle Damage | Cars, bikes, scooters, trucks |
| 🏠 Property Damage | Walls, roofs, windows, buildings |
| 🪑 Household Items | Furniture, appliances, glass items |
| 🚚 Logistics Claims | Courier and shipping damages |
- Upload claim images
- Multiple claim categories
- Image preview
- Evidence validation
- Claim description
- Real-time AI processing
- Damage detection
- Severity estimation
- Object recognition
- Image quality assessment
- AI-generated reasoning
- Confidence scoring
- Evidence consistency validation
- Historical claim verification
- Risk categorization
- Suspicious claim detection
- Manual review recommendations
- Total claims
- Approval trends
- Rejection statistics
- Severity distribution
- Interactive charts
- Business insights
Generate professional reports containing:
- Claim summary
- Uploaded evidence
- AI findings
- Fraud assessment
- Confidence score
- Final recommendation
Every decision includes:
- AI reasoning
- Confidence score
- Evidence explanation
- Decision transparency
- Recommended next steps
User Uploads Image
│
▼
Streamlit Application
│
▼
Claim Information Input
│
▼
Image Quality Validation
│
▼
Computer Vision (OpenCV)
│
▼
Google Gemini AI
│
▼
Evidence Verification
│
▼
Fraud Risk Assessment
│
▼
Decision Engine
│ │
▼ ▼
Analytics Dashboard PDF Report
flowchart TD
A[Upload Damage Image]
B[Claim Information]
C[Image Validation]
D[Damage Detection]
E[Gemini AI Analysis]
F[Evidence Verification]
G[Fraud Assessment]
H[Decision Engine]
I[Analytics Dashboard]
J[Generate PDF Report]
A --> B
B --> C
C --> D
D --> E
E --> F
F --> G
G --> H
H --> I
H --> J
ClaimVision-AI/
│
├── app.py
├── main.py
├── report_generator.py
├── requirements.txt
├── README.md
│
├── src/
│ ├── claim_extractor.py
│ ├── image_analyzer.py
│ ├── evidence_checker.py
│ ├── history_checker.py
│ └── decision_engine.py
│
├── models/
├── notebooks/
├── evaluation/
├── reports/
├── utils/
└── assets/
| Module | Responsibility |
|---|---|
| app.py | Streamlit interface |
| claim_extractor.py | Collects and structures claim data |
| image_analyzer.py | Damage detection using OpenCV |
| evidence_checker.py | Validates uploaded evidence |
| history_checker.py | Performs fraud-related checks |
| decision_engine.py | Generates final recommendation |
| report_generator.py | Creates downloadable PDF reports |
| Layer | Technologies |
|---|---|
| Frontend | Streamlit, HTML, CSS |
| Backend | Python |
| AI | Google Gemini AI |
| Computer Vision | OpenCV |
| Machine Learning | Scikit-Learn |
| Data Processing | Pandas, NumPy |
| Visualization | Plotly |
| Reporting | ReportLab |
git clone https://github.com/Komal2008/ClaimVision-AI.git
cd ClaimVision-AIpython -m venv venv
venv\Scripts\activatepython3 -m venv venv
source venv/bin/activatepip install -r requirements.txtCreate a .env file:
GEMINI_API_KEY=YOUR_API_KEYstreamlit run app.py| Step | Description |
|---|---|
| 1 | Upload damage image |
| 2 | Select claim category |
| 3 | Enter claim details |
| 4 | AI analyzes evidence |
| 5 | Validate evidence |
| 6 | Assess fraud risk |
| 7 | Generate recommendation |
| 8 | View dashboard |
| 9 | Export PDF report |
- Multi-image verification
- OCR for claim documents
- Video damage analysis
- RAG-based insurance assistant
- REST API
- Database integration
- User authentication
- Cloud deployment
- Docker support
- CI/CD pipeline
- Unit & integration testing
Contributions are welcome and appreciated!
- Fork the repository.
- Clone your fork.
- Create a feature branch.
- Make your changes.
- Test your implementation.
- Commit with a meaningful message.
- Push your branch.
- Open a Pull Request.
feature/add-dashboard
bugfix/fix-image-upload
docs/update-readme
feat: add fraud risk module
fix: resolve image preprocessing bug
docs: improve installation guide
refactor: simplify decision engine
- Code builds successfully
- Existing functionality is not broken
- Documentation updated
- New feature tested
- Clean commit history
- Follows project structure
Please include:
- Python version
- Operating system
- Steps to reproduce
- Expected behavior
- Actual behavior
- Screenshots (if available)
- Insurance claim verification
- Package damage assessment
- Electronics damage inspection
- Property damage review
- Vehicle damage analysis
- Logistics evidence verification
- AI-assisted claim processing
- Fraud detection
This project is licensed under the MIT License.
Engineering Student | AI & Machine Learning Enthusiast | Open Source Contributor
- GitHub: https://github.com/Komal2008
If this project helped you or inspired your work:
⭐ Star this repository
🍴 Fork it
🛠️ Contribute
📢 Share it with the community