That is the approximate number of airports and airfields around the world. To monitor them all would be a herculean task. But, what if we could monitor them all?
Last month, CrowdAI announced a computer vision (CV) capability that provides not just the location, time, and quantities of aircraft, but identifies aircraft by type, e.g. FELON, J-20, F-4, etc. This tool offers an unprecedented look at foreign aircraft holdings, known as Air Order of Battle. But, it begged the question: once we had the AI model, what did we do with it? Some background, first: Unbeknownst to many, geospatial intelligence, or GEOINT, requires significant plumbing to move pixels from “sensor-to-glass,” as they say. Once that data is on Earth and processed for use, an analyst must perform multiple, discrete tasks to search, retrieve, and analyze the images, which are separate from any reporting and dissemination of their analyses.
Those rote tasks consume an imagery analyst’s day. AI can now handle exploitation, but that doesn’t resolve the rest of the workflow or answer how to plumb which satellites to which models to which end-users. What’s more, the massive influx of Earth observation satellites has only compounded this problem. When commercial satellites were counted by the onesies and twosies, it was far simpler to route data. Today, scores of potential sensors have to be plumbed correctly. Now, repeat that process thousands of times over for every airfield. Or, do as we have done and automate.
(Graphic 1) Persistent Analytic Pipelines are established quickly and easily using the CrowdAI platform (left graphic). Areas of interest are drawn graphically, which then query a variety of imagery providers, e.g. Global EGD, BlackSky, Planet, etc. Then, the imagery is filtered to the appropriate computer vision model based on sensor, mission (air, ground, navy, etc.), and biome. Model detections are disseminated to a designated repository or dashboard (right graphic).
In our no-code platform, CrowdAI has added what we call Persistent Analytic PipelinesTM. This feature enables users to automate the end-to-end GEOINT analysis workflow, known as Processing, Exploitation, and Dissemination. (We’ll tackle tasking and collection another day). Seamlessly, each persistent pipeline connects repositories of satellite imagery (e.g. Maxar, Blacksky, and Planet, etc.) to corresponding CV models, and, then, to an intelligence database. Multiple models can be connected to multiple imagery sources, supporting advanced spatio-temporal analysis over any airfield (or different target using a different CV model, e.g. ships in maritime ports).
Today, our AOB system processes imagery over dozens of airfields spanning an entire continent, using 17 satellites (Graphic 2). Those 17 discrete data feeds are automatically processed through CV pipelines. As a fused analytic, this hybrid constellation of commercial satellites provides excellent target revisit for pattern of life analysis and maintaining target custody.
Because the tool places imagery queries based on geographic location, any image footprint that intersects with the target area is automatically pulled through the pipeline, generating a GEOINT analytic result. With CV and persistent analytic pipelines feeding a dashboard or common operating picture (Graphic 3), a single analyst can monitor air activity across an entire country, or even multiple countries, depending on the volume of traffic.
(Graphic 2) Today, CrowdAI’s worldwide analytic service for airport monitoring is fed by 17 satellites, providing both spatial resolution for precision aircraft identification as well as rapid airport revisit for aircraft custody. Persistent Analytic Pipelines ingest multiple data streams and generate clean, structured observations for spatio-temporal analysis.
Now imagine similar pipelines set up for navy and ground orders of battle: every ship in every port identified by class and accounted for; ground units found bivouacked in the field; analysts empowered by individualized dashboards, allowing them to cut, slice, and geospatialize data to meet their unique needs. Imagine a "Wall of Knowledge" fed by hundreds or thousands of pipelines, automatically revealing GEOINT-derived insights around the globe moments after being captured on imagery. That future is here.
(Graphic 3) An example airport monitoring dashboard over Ukraine. Similarly, a “Wall of Knowledge” can be fed by hundreds of persistent analytic pipelines monitoring air, ground, navy and other missions as selectable layers.
With CrowdAI, an intelligence officer monitoring a massive GEOINT target deck can focus on the why and what’s next, while persistent analytic pipelines and AI do the work of what, where, and when.
To learn more about this game-changing technology and how it can support your mission, please contact firstname.lastname@example.org.