Ingest EO or IR video
The suite accepts live video from gimbals, observation posts, UAV payloads or STANAG 4609 streams over RTSP, UDP/RTP or direct camera inputs.
Perception
Video Analytics Suite
Real-Time Video Intelligence at the Tactical Edge
Overview
What the Video Analytics Suite does
Edge-AI video analytics for automatic detection, classification and persistent tracking from EO and IR video.
The Neuronic Video Analytics Suite is an edge-AI automatic target recognition (ATR) and detector-tracker that runs onboard drones, ground robots, observation posts and vehicle sensor suites. It detects, classifies, tracks and re-identifies personnel, vehicles and small aerial targets in EO and IR video, and publishes target metadata to C2, gimbal control and autonomy without a datalink to a ground processing centre.
Embedded ATR detector-tracker that runs onboard drones, UGVs, observation posts and vehicle sensors — turning EO/IR video into target data at the edge, with no datalink to a ground station.
Detects, classifies, tracks and re-identifies targets in real time, holds lock through occlusion and platform motion, and feeds autonomy, gimbal and C2. Keeps working when the link is degraded or denied.
Works with existing sensors and third-party detectors and publishes metadata over standard defense interfaces — drops into current ISR and C-UAS architectures.
Capabilities
Core capabilities
- Automatic Target Recognition (ATR)
- Detection and classification of personnel, vehicles and aerial targets
- Multi-target tracking with persistent track IDs
- Re-identification after occlusion or field-of-view loss
- Area-of-interest (AOI) precision tracking
- EO and IR processing, day and night
- Third-party detector and model integration
- Target metadata and reporting to C2 and autonomy
- STANAG 4609 / VMTI metadata integration where applicable
- Counter-UAS detection and tracking of small aerial targets
How it works
How the Video Analytics Suite works
From sensor input to actionable output in four steps.
Detect, classify and track
Onboard models locate targets, assign class labels and give each target a persistent track ID that survives crossing, grouping and temporary occlusion.
Hold precision AOI tracking
The operator selects a target or area of interest; the tracker keeps it centred and can drive the gimbal or cue other systems.
Publish structured metadata
Target position, class and track information flow to C2, autonomy and gimbal control over STANAG 4609 / VMTI, JSON, ROS or MAVLink.
Use cases
Where it's used
Typical situations this system is deployed in, and what it solves in each one.
Watching a large area for hours
Nobody can stare at a video feed all day. The software watches instead, and reports people and vehicles as they appear, from a drone or a fixed observation post.
Spotting small drones
Small aerial targets are hard to see against a cluttered background. They are detected, classified and tracked, and the track can be handed to the systems that respond.
Base and perimeter watch
Continuous automated monitoring day and night, with EO and IR video, so an alert only reaches the operator when something matters.
Keeping a target in view
Once a target is marked it stays tracked through movement and brief loss of sight, and the position can be passed to a gimbal, a drone or a ground robot.
Focus area
Counter-UAS
- Detection and classification of small, low-signature aerial targets
- Persistent tracking through cluttered and low-contrast backgrounds
- EO/IR fusion-ready pipeline for day and night engagement support
- Track hand-off to effectors, gimbals and command-and-control
Integration
Interfaces and integration
Built to drop into an existing architecture — standard interfaces, standard hardware, no forced replacement of fielded systems.
- Video in: RTSP, UDP/RTP, MIPI/CSI, USB, HDMI capture, STANAG 4609 streams
- Metadata out: STANAG 4609 / VMTI, JSON over TCP/UDP, ROS topics, MAVLink target messages
- Runs on NVIDIA Jetson Orin class hardware; also deployable on x86 rugged computers
- Model updates and class sets adapted to customer sensors and mission profiles
- Cross-cueing into NOMAD Tactical Robotic C2 and third-party C2 including ATAK
Related
Works with
FAQ
Common questions about the Video Analytics Suite
Plain-language answers for integrators, procurement teams and operators.
- What is automatic target recognition (ATR)?
- ATR turns raw EO or IR video into detected, classified and tracked targets. Instead of an operator watching every frame, the software produces a structured target list with persistent identities that can be passed to C2, effectors or autonomy.
- Does the processing happen at the edge or in the cloud?
- All processing runs on the onboard edge computer, typically NVIDIA Jetson Orin class hardware. This keeps the capability alive when bandwidth is limited, jammed or denied.
- Can it operate at night or in poor visibility?
- Yes. The pipeline processes both EO and IR channels, so detection and tracking continue after dark and in conditions where EO alone is insufficient.
- What target types can it detect?
- Personnel, vehicles and small aerial targets are supported out of the box. Class sets can be adapted to customer sensors and mission profiles.
- How does it integrate with my gimbal or C2 system?
- Target metadata is published over STANAG 4609 / VMTI, JSON, ROS topics or MAVLink target messages. The suite also cross-cues into NOMAD and third-party C2 including ATAK.
- Is it suitable for counter-UAS missions?
- Yes. The suite detects, classifies and tracks small, low-signature aerial targets and can hand tracks off to effectors, gimbals or command-and-control systems.
Discuss your platform and mission
Tell us the platform, sensor and mission. We will walk you through integration options and arrange a technical demonstration.