ATC-45 compliant post-disaster safety evaluation for ATAK — now with AI-powered damage detection from drone imagery. Rapid and detailed building inspections with standardized posting categories, from earthquake response to hurricane damage documentation.
Revolutionary on-device AI analysis transforms post-disaster drone flights into actionable intelligence. Process orthomosaic imagery (GeoTIFF/VRT) through computer vision and vision-language models to automatically detect and prioritize damaged buildings for field inspection.
Run MiniCPM-V, Gemma 3N Vision, or Phi-4 entirely offline. No cloud connectivity required — critical for disaster zones with damaged infrastructure. AI analyzes building tiles for structural damage, fire damage, debris patterns, and generates preliminary posting recommendations.
| Stage | Technology | Output |
|---|---|---|
| 1. Imagery Loading | Rasterio/NumPy | Orthomosaic tiles per building |
| 2. Footprint Intersection | OSM/Microsoft Building Data | Building polygons with IDs |
| 3. CV Detection | OpenCV via Chaquopy | Damage indicators (rubble, fire, cracks) |
| 4. LLM Analysis | MiniCPM-V / Gemma 3N Vision | Detailed damage descriptions |
| 5. Priority Queue | Damage Score Algorithm | Inspection queue by severity |
| 6. ATC-45 Pre-fill | Java Service | Pre-populated field forms |
AI-generated damage scores map directly to ATC-45 posting categories:
| Damage Score | Posting Category | Priority |
|---|---|---|
| < 30% | 🟩 INSPECTED (Green) | Lower priority |
| 30% - 70% | 🟨 RESTRICTED USE (Yellow) | Medium priority |
| > 70% | 🟤 UNSAFE (Red) | Immediate inspection |
The Building Damage Assessment module transforms ATAK into a powerful tool for structural safety evaluations following earthquakes, storms, floods, or other disasters. Built on ATC-45 standards from the Applied Technology Council, it provides standardized methodology for trained inspectors to assess building safety and communicate results.
When disaster strikes, every minute counts. Start a Rapid Evaluation in seconds, capture location with GNSS, document damage with photos, and post your assessment — all offline, all in the field. Results sync when connectivity is restored.
Each assessment type serves a different phase of post-disaster evaluation:
| Assessment | Duration | When to Use |
|---|---|---|
| Rapid Evaluation | 15-30 minutes | First-pass safety screening of large areas |
| Detailed Evaluation | 1-4 hours | Comprehensive structural review of specific buildings |
| Equipment Assessment | 30-60 minutes | Fixed equipment safety (HVAC, generators, elevators) |
Every assessment concludes with a posting recommendation. These placard categories provide immediate visual communication of building safety:
| Category | Placard | Meaning |
|---|---|---|
| INSPECTED | 🟩 Green | Building appears safe for occupancy |
| RESTRICTED USE | 🟨 Yellow | Limited entry with specific restrictions |
| UNSAFE | 🟤 Red | Do not enter — structural hazard |
Comprehensive forms capture all required ATC-45 evaluation data:
Attach multiple geotagged, timestamped photos to each assessment. Visual documentation is critical for damage evidence and supports the posting decision.
Each assessment can include unlimited photos. The system captures location and timestamp automatically. Photos appear in generated reports, providing a complete evidence package for decision-makers.
Three methods for precise building location:
| Method | Accuracy | When to Use |
|---|---|---|
| GNSS | Meters to centimeters | Standing at the building |
| Map Tap | Map dependent | Remote assessment or planning |
| Manual Entry | Exact | Known coordinates from survey |
All assessments are stored locally in OGC GeoPackage format with spatial indexing. This enables:
Generate professional reports from any assessment:
| Report Type | Contents |
|---|---|
| Assessment Report | Complete form data, location map, photos, posting recommendation |
| Event Summary | Statistics by posting category, affected area overview, inspector activity |
Find patterns in damage distribution:
Use the spatial query feature to find all assessments within a radius. Identify clusters of UNSAFE postings that may indicate localized ground failure or structural deficiencies in a specific construction type.
Built for field use in disaster zones where connectivity is unreliable or unavailable. AI damage detection runs entirely on-device with no cloud dependencies: