Capability
Classification & Context
State becomes useful when context is visible.
Evidence levelE0Concept
The functional gap
Knowing that something is there is not knowing what it is. A label stated too early becomes a fact in the next system, and a missing label is not an empty class.
From inputs to outputs
Takes in
- Tracks with their state and confidence
- Optical or other cues attached to a track
- Context: place, time and conditions
Does
Offers a class for a track together with the cues behind it, and keeps classification apart from identity. An unresolved track stays unknown.
Hands on
- A class or an explicit unknown
- The cues that support the class
- Context shown beside the class, not folded into it
A worked example
Three steps from a synthetic scenario, in relative terms.
| Step | What is seen | What is stated |
|---|---|---|
| 1 | A track has a slow, steady path | A class is offered with its cues |
| 2 | The cues are weak or conflicting | The track is shown as unknown |
| 3 | A clearer cue arrives later | The class changes and the change is recorded |
| Item | Share |
|---|---|
| Class A | 38 |
| Class B | 29 |
| Class C | 21 |
| Class D | 12 |
Confidence and uncertainty in this function
A class carries its own confidence, separate from the confidence in the track. A confident track can still have an open class.
Evidence and maturity
This page claims nothing above the level of the projects it rests on.
- Electro-Optical AI Detection & ClassificationEvidence level: E0 · Concept
- Drone Detection & Airspace Intrusion IntelligenceEvidence level: E0 · Concept
- Bird & Wildlife Radar IntelligenceEvidence level: E0 · Concept
Where it applies
- Critical InfrastructureEarlier, evidence-backed airspace awareness around important sites.
- Airports & AviationUnderstand the airspace around active aviation operations.
- Low-Altitude & UAS AwarenessUnderstand low-altitude activity across multiple sources.
- Wildlife & Aviation SafetySee wildlife activity in the context of aviation operations.
What this does not claim
- Classification is not identification; no page here names an operator or an intent.
- Class labels describe the design and state no hit rate or error rate.
- A class is revisited when better cues arrive, and the earlier state stays on record.
Projects behind this
- Electro-Optical AI Detection & ClassificationSee the observation. Understand the evidence. Keep uncertainty visible.
- Drone Detection & Airspace Intrusion IntelligenceDetect activity. Build context. Preserve the evidence behind the airspace state.
- Bird & Wildlife Radar IntelligenceUnderstand movement patterns in low-altitude airspace, not just individual detections.