Different environments.
Specific intelligence.
Build recognition around the objects and workflows that matter in your sector. Start with a defined problem. Develop the model layers to solve it.
Turn inspection imagery
into asset intelligence.
Build a structured view of infrastructure from visual inputs. Recognize equipment classes and components, support condition review, and connect observations to asset-management workflows.
Discuss a utility audit- asset_class
- power transformer
- component
- bushing · radiator · tap changer
- condition
- flagged for review
- location
- site · bay · source frame
- evidence
- frame hash · reviewer decision
Know what is present.
Understand the context.
Recognize customer-defined vehicle and equipment classes, structure visual inventories and support operational review across yards, facilities and logistics environments.
Discuss asset recognition- vehicle_class
- customer-defined
- attributes
- trailer · container · plate region
- activity
- arrival · dwell · departure
- evidence
- track · source frames · reviewer
Your recognition vocabulary.
Engineered into the system.
Define the equipment families, aircraft categories and visual attributes your application needs. Custom model layers bring specialized vocabularies into asset inventory and inspection workflows.
Discuss a custom model- aircraft_class
- customer-defined
- attributes
- configuration · registration region
- status
- parked · towed · in service
- evidence
- frame hash · reviewer decision
Threats seen early.
Decisions made by people.
Multi-class threat detection on recorded and uploaded footage: firearms, edged and blunt weapons, explosives, and fire and smoke, with person and tool context to separate genuine threats from everyday look-alikes.
Candidates are verified at original resolution and confirmed by a reviewer, and every observation keeps a hashed, auditable evidence trail from upload to resolution.
- threat_class
- firearm · edged weapon · fire and smoke
- context
- person · tool
- verification
- corroborated · uncertain · skipped
- decision
- reviewer · confirmed or dismissed
Vessels, vehicles
and patterns of life.
Recognize vessel and vehicle classes from coastal cameras, aircraft and satellite imagery, and turn repeated observations into patterns of activity.
Model layers are built around your sensors and the distinctions your analysts care about, to cut watch-floor workload and shorten the time from detection to response.
Discuss maritime recognition- vessel_class
- customer-defined
- pattern
- repeated observation · route · dwell
- sensor
- coastal camera · aircraft · satellite
- evidence
- track · source frames · analyst decision
Weapons, platforms
and units. Identified.
Platform and weapon-system identification, unit and formation indicators, equipment inventories and readiness workflows, built for drone, aircraft and satellite imagery in contested conditions.
Vektra develops model layers around customer-defined requirements, with edge deployment for denied environments and controlled infrastructure.
Explore defense recognition- platform
- class · type · variant
- unit_indicators
- markings · equipment mix · count
- confidence
- calibrated · unknown class · decoy likelihood
- decision
- human · recorded apart from the evidence