Evaluation Criteria
How your solution will be evaluated
Evaluation will consider innovation, scalability, performance, usability, feasibility, and real-world deployment potential. UI/UX also plays an important role. Real world compatibility with government infrastructure and state-scale readiness are important considerations.
Model-Specific Rubrics
Each model has specific evaluation criteria based on its objectives and requirements:
Model 1 CCTV Registry & GIS Mapping
| Evaluation Criteria | Marks | Description / Justification |
|---|---|---|
| Data Completeness & Accuracy | 20 | All required fields captured; GPS coordinates validated; no missing critical data |
| GIS Integration & Mapping Quality | 20 | Accurate map rendering; layers for zones/districts; real-time location updates |
| Search & Filter Functionality | 15 | Multi-parameter search; filter by zone, status, type; fast query response |
| Scalability to 1,00,000+ Records | 15 | Database optimized for large datasets; pagination and lazy loading implemented |
| UI/UX & Dashboard Design | 15 | Intuitive interface; responsive design; clear visual hierarchy |
| API Design & Documentation | 10 | RESTful APIs; proper versioning; Swagger/OpenAPI docs provided |
| Security & Access Control | 5 | Role-based access; encrypted data transmission; audit logs |
| Total | 100 |
Model 2 Stream Unification & Vendor Agnostic Middleware
| Evaluation Criteria | Marks | Description / Justification |
|---|---|---|
| Protocol Support (RTSP, ONVIF, HLS etc.) | 20 | Supports multiple streaming protocols; vendor-agnostic integration demonstrated |
| Stream Stability & Latency | 20 | Low-latency streaming; auto-reconnect on failure; buffer management |
| Middleware Architecture | 20 | Clean abstraction layer; modular design; easy to add new camera vendors |
| Concurrent Stream Handling | 15 | Handles 100+ simultaneous streams without degradation |
| Transcoding & Format Normalization | 10 | Converts streams to unified format; quality preservation |
| Fault Tolerance & Logging | 10 | Graceful error handling; detailed logs; alerting on stream failure |
| Security & Authentication | 5 | Secure stream access; token-based auth; encrypted transport |
| Total | 100 |
Model 3 Centralized Monitoring & Control Room
| Evaluation Criteria | Marks | Description / Justification |
|---|---|---|
| Live Feed Display & Grid Layout | 20 | Multi-feed grid view; customizable layouts; full-screen mode support |
| Alert & Incident Management | 20 | Real-time alerts; incident logging; escalation workflow |
| PTZ & Camera Control | 15 | Pan/tilt/zoom controls; preset positions; tour scheduling |
| Recording & Playback | 15 | Scheduled/event-based recording; timeline scrubbing; export functionality |
| Performance Under Load | 15 | Dashboard loads under 3 seconds; 500+ concurrent users supported |
| UI/UX & Operator Usability | 10 | Minimal training required; keyboard shortcuts; dark mode for control rooms |
| Audit Trail & Reporting | 5 | User action logs; shift reports; exportable incident summaries |
| Total | 100 |
Model 4 Centralized Analytics & AI Insights
| Evaluation Criteria | Marks | Description / Justification |
|---|---|---|
| AI/ML Model Accuracy | 25 | Detection accuracy above 90%; low false positive rate; tested on real-world data |
| Analytics Features (crowd, behaviour, ANPR) | 20 | Breadth of analytics; crowd density, anomaly detection, license plate recognition |
| Real-time Processing Speed | 20 | Inference latency under 200ms; GPU/edge optimization demonstrated |
| Dashboard & Visualization | 15 | Heatmaps, trend charts, incident timelines; exportable reports |
| Scalability & Edge Deployment | 10 | Works on edge devices; scales to state-level camera count |
| Model Explainability & Bias | 5 | Explainable AI outputs; bias mitigation documented |
| Integration with Other Models | 5 | APIs compatible with Model 1/2/3; unified data pipeline |
| Total | 100 |
Performance Benchmarks
Your solution should meet the following performance targets:
| Metric | Target Value |
|---|---|
| Camera Records / Endpoints Support | 1,00,000+ |
| API Response Latency | Under 200 ms |
| Dashboard Load Time | Under 3 seconds |
| Detection / Processing Accuracy | Above 90% |
| System Uptime Target | Above 99% |
| Concurrent Users (without degradation) | 500+ |
Bonus Criteria
Additional weightage may be given for:
- Innovation beyond the defined scope
- AI and machine learning integration
- Scalability for state-level deployment
- Optimization for low-bandwidth environments
- User-centric design
Global Judging Themes
Technical Soundness
Quality of code architecture and implementation
Scalability
Ability to handle growth and increased load
Security & Access Control
Authentication, authorization, and data protection
UI/UX & Usability
User interface quality and user experience
Performance & Latency
Speed and responsiveness of the system
Feasibility for Government
Readiness for real-world deployment
Innovation & Value Addition
Novel approaches and unique contributions
AI/ML Intelligence
Machine learning integration where relevant