QGIS 4.x Plugin & ArcGIS Pro Add-In · MISB ST 0601 · YOLOv11 · No proprietary licenses required
Intelligence teams spend hours scrubbing through video to find targets. Existing tools require expensive extensions, are tied to specific platforms, or lack AI automation entirely.
ArcGIS Image Analyst add-on costs thousands per seat — often for analysts who only need FMV.
Frame-by-frame scrubbing to spot objects of interest. No automation, no AI, no workflow.
Classified or contested environments cannot send video to cloud AI services. Air-gap is mandatory.
A dual-platform add-on that brings automated AI detection, live map sync, and structured analyst review to both QGIS 4.x and ArcGIS Pro — without Image Analyst.
Pure Python + PyQt6 + PyQGIS. Installs from the QGIS Plugin Manager. Cross-platform: Windows, macOS, Linux.
Python 3.12 · PyQt6 · PyQGIS · OpenCV-Python · ONNX Runtime (GPU) · FFmpeg subprocess
GeofmvDockWidget · KlvParser · FootprintLayerManager · OnnxDetector · OpenRouterClient
QgisAuthManager for API keys · ONNX Runtime DirectML on Windows · VideoCapture frame extraction
QGIS Plugin Repository ZIP · Git clone install · No compiled binaries required
C# / .NET 8 / ArcGIS Pro SDK 3.5. Windows only. Distributed as a signed .esriAddinX package. Does not require the Image Analyst extension.
C# 12 · .NET 8 · ArcGIS Pro SDK 3.5 · LibVLCSharp · OpenCvSharp4 · Microsoft.ML.OnnxRuntime
GeofmvDockPane · KlvParser · FootprintFeatureClassManager · OnnxDetector
LibVLCSharp replaces VideoLayer. Handles MPEG-TS, H.264/H.265, RTSP streams.
Windows Credential Manager via Windows.Security.Credentials.PasswordVault for API keys.
Every extracted frame travels through the same 5-step pipeline regardless of which AI mode is active.
YOLOv11 / custom ONNX models · DirectML (Windows) · CUDA (Linux) · ~30ms/frame · Air-gap safe
QWEN3-VL · Llama-3.2-Vision · ERNIE 4.5 · GPT-4.1-mini · Base64 frame encoding · Falls back to local
| Standard | Coverage | Notes |
|---|---|---|
| STANAG 4609 | ✔ Full | MPEG-TS container, H.264/H.265, KLV metadata mux |
| MISB ST 0601 | ✔ 146 tags | Round-trip fidelity, BER length, checksum verification |
| MISB ST 0102 | ✔ Partial | Security metadata preservation (read/pass-through) |
| MISB EG 0104 | ✔ Compliant | Predator UAV compatibility |
| MISB ST 0903 | ✔ Full | VMTI export with VObject/VTrack series |
| OGC Moving Features | ✔ GeoJSON-MF | Trajectory export format |
An embedded OLAP engine in the browser. Query detection archives, spatial data, and cloud storage without a server.
ST_Read, GeoParquet, GeoJSON, FlatGeobuf, WMS layers
S3, Azure Blob, GCS, HTTPS with range-request support
ZIP archives, KMZ, 3TZ 3D Tiles metadata exploration
GeoPackage, Survey123, Fulcrum field-collection databases
PostGIS, Supabase, ArcGIS Enterprise SDE joins
Apache Iceberg time-travel queries on S3/ADLS detection archives
| Feature | QGIS 4.x | ArcGIS Pro |
|---|---|---|
| STANAG 4609 / KLV parsing | ✔ | ✔ |
| Live map footprint sync (5 Hz) | ✔ | ✔ |
| Local ONNX AI inference | ✔ | ✔ |
| Remote OpenRouter AI | ✔ | ✔ |
| DJI telemetry mux → STANAG | ✔ | ✔ |
| Analyst review workflow | ✔ | ✔ |
| VMTI ST 0903 export | ✔ | ✔ |
| Image Analyst license required | ✘ | ✘ |
Every AI detection bounding box is converted to a geographic point using a two-stage algorithm.
4+ Ground Control Points extracted from KLV tags. OpenCV findHomography() maps pixel coords to world space. Applied when GCPs are available and spread.
Slant range from Tag 21 + sensor angles (Tags 17/18/10/11) project a ray from sensor altitude to ground plane. CEP ≤ 15 m.
| Format | Use Case | Analyst Filter |
|---|---|---|
| GeoPackage (.gpkg) | Primary QGIS/ArcGIS exchange | ✔ Status, Class, Confidence, Time |
| GeoJSON (.geojson) | Web mapping, REST APIs | ✔ Status, Class |
| KML (.kml) | Google Earth, legacy viewers | ✔ Class, Time range |
| Shapefile (.shp) | Legacy GIS compatibility | ✔ Status |
| VMTI ST 0903 | NATO FMV tools, mission systems | ✔ Accepted detections only |
| GeoTIFF Mosaic | Orthorectified scene reconstruction | – |
Exploits STANAG 4609 feeds from UAVs. Uses AI detection + review workflow to locate targets 10× faster than manual scrubbing.
Ingests DJI drone footage for infrastructure inspection. Uses DuckDB analytics to query detections against asset databases.
Situational awareness from body-cam or aerial video during emergencies. Exports KML/GeoPackage for command dashboards.
Plugins → Manage → Search "GeoFMV" → Install. Or: install from ZIP file (Plugins → Install from ZIP).
Project → Add-In Manager → Add From File → select GeoFMV.esriAddinX. Restart ArcGIS Pro.
Click the GeoFMV toolbar button. Drag and drop any STANAG 4609 .ts file — or a DJI MP4 + CSV pair — and press Play.