SMARTIONIXMartion Mobile
A person can start an eligible assistance request and provide available location and configured SOS context.
Martion PIOS · Public safety acquisition
See the signal. Review the context. Respond.
Martion connects user-started citizen requests, activated camera intelligence and authorised responder screens in a controlled operational workflow.
Complete operating chain
Three surfaces. One event lifecycle.
Martion separates the citizen, camera and command-room roles while keeping their eligible events connected.
Mac Martion Vision
When explicitly activated, eligible local or configured IP camera feeds can be analysed for defined event candidates.
Martion TV Responder
An authorised Apple TV installation receives scoped events with available map, assignment and resolution controls.
From camera to responder
Camera context becomes an operational signal.
A configured event is evaluated on the Mac. When its conditions are met, Martion sends the event type, confidence or match context, camera identity and available location to the authorised responder workflow.
- 01Camera sourceBegin with the Mac camera for a demonstration, then evaluate supported IP camera feeds for deployment.
- 02Local candidate analysisConfigured activity and identity-reference checks run on the authorised Mac session.
- 03Scoped alert deliveryThe event is delivered to the registered responder role assigned to the workflow.
- 04Live reviewThe responder can inspect the available context and, where enabled, request the related live camera view.
POLICE VISION / CAMERA 01ACTIVE
CANDIDATE EVENT
Possible knife detected
Candidate score and camera location attached for review
RESPONDER
Police Responder · Office
Alert received in the scoped live inbox
ACTIONS
Review feed · Assign · Resolve
Available controls depend on responder role
Edge AI architecture
Analyse locally. Escalate the signal that matters.
Martion Vision is designed to keep candidate analysis close to the authorised camera source while the connected service coordinates scoped alerts, responder state and live review across locations.
Local candidate analysis
Eligible camera frames are evaluated on the authorised Mac instead of requiring every continuous frame to become a cloud record.
Purpose-limited escalation
A qualifying event can carry the event type, source identity, location and authorised review context to the responder workflow.
Registered expansion
Compatible cameras and sites can be introduced through defined source identity, location, event rules and operational ownership.
Human verification
Edge AI supports earlier awareness. Trained personnel remain responsible for verification, response and resolution.
Local inference can reduce data movement and event latency. Remote delivery, live review and cross-site synchronisation still depend on secure network availability.
Operational capabilities
Signals designed for authorised review.
Each capability requires a defined operating purpose, configured source and acceptance test. Outputs support assessment; they do not replace investigation or human judgement.
Close-camera activity
Configured near-camera conditions can identify a person, face or hand approximately 30–45 cm from an eligible camera.
Possible knife detection
Classifier confidence and blade-form checks can create a candidate alert above the configured threshold for responder review.
Missing or wanted person
Authorised iCloud reference folders can support candidate comparisons, with the reference filename supplying the displayed name.
Vehicle plate candidates
Moving plate text can be read and compared with an authorised stolen-vehicle reference folder.
Location and camera access
Eligible alerts include the camera identity, available location and a controlled action to inspect the live feed.
Public-safety broadcasts
Authorised police can target a supported radius, road corridor or polygon and receive aggregate delivery counts.
Drone and drone-swarm candidates
Invisible Shield searches configured camera feeds for possible single or multiple UAS targets, retains fast-moving candidates through predictive visual tracking and sends source-specific alerts for responder verification.
Financial investigation
Trace connected accounts from statements to the terminal beneficiary.
Martion AI Investigator helps authorised teams convert supplied bank statements into a reviewable fund-flow map. It links incoming accounts, hopping accounts, shared destinations and outward movement while preserving each conclusion’s source transaction.
- 01PDF and CSV ingestionNormalise UPI, IMPS, RTGS, NEFT, card, cash and remittance entries from supplied records.
- 02Interconnected fund flowsShow credits, supplied accounts and debit destinations in one navigable evidence graph.
- 03Pattern-led reviewSurface shared destinations, possible layering, standalone test payments and foreign-remittance indicators for human assessment.
- 04Evidence-ready reportingProduce a concise report with source references, account rankings, confidence limits and recommended investigative follow-up.
FUND FLOW / EVIDENCE GRAPHREVIEW
SOURCES
Originating accounts
Names, transaction rails, dates, amounts and statement references
HOPS
Shared and intermediary accounts
Connections across the supplied statements and likely onward movement
TERMINAL REVIEW
Ultimate-beneficiary candidates
Ranked by traced value retained at the end of observed chains
Responder categories
The right controls for each role.
Martion TV separates operational categories so a station or team receives only the event types and actions assigned to its work.
Police Responder
Citizen assistance, assignment, event-bound tracking and Police Vision review.
Traffic Police
Traffic Mode approval, emergency corridors and road-focused public notices.
Anti Terrorist Squad
A separately registered category for authorised operations and assigned events.
Emergency response
Eligible fire, rescue and general-distress events with mapped incident context.
Evaluation progress
Demonstration completed. Pilot planning underway.
Martion’s public-safety pathway is being advanced through direct demonstrations and controlled pilot planning with police organisations.
CompletedHyderabad · Telangana
02
Successful demonstrations with Hyderabad Police
Smartionix has completed two successful Martion demonstrations with Hyderabad Police, presenting the connected citizen, Vision and responder workflow for public-safety evaluation.
Planning in progressJammu & Kashmir
PILOT
Ongoing pilot planning with Jammu & Kashmir Police
Smartionix is planning a controlled Martion pilot with Jammu & Kashmir Police to define operating scope, participating locations, responder roles, acceptance tests and field evaluation.
These milestones describe evaluation activity. Any operational deployment remains subject to agency approval, agreed pilot terms and field-acceptance criteria.
Acquisition pathway
Evaluate before you scale.
A controlled pilot makes accuracy, latency, permissions, connectivity and responder workflow measurable in the intended field conditions.
Operational scope
Define locations, responder roles, camera sources, event categories and legal or policy requirements.
Controlled setup
Register the responder screen, configure selected cameras and establish the scoped service connection.
Field acceptance
Measure event conditions, false alerts, missed events, latency, live review and recovery behaviour.
Phased rollout
Review evidence, refine thresholds and agree whether the system is ready for additional sites.
Responsible operation
Human judgement stays central.
Computer vision can be uncertain. Lighting, angle, motion, occlusion, image quality and lookalike objects affect results. Martion therefore treats detections as candidates for authorised review.
- 01Explicit activationCamera analysis begins only in an activated, configured session.
- 02Candidate, then verificationA system score triggers review; it does not establish guilt, identity or intent.
- 03Controlled referencesMissing, wanted-person and stolen-vehicle folders must be maintained by the authorised organisation.
- 04Defined retentionLive review and connected records follow the controls described in the Martion privacy policy.
False-alert rate
Test ordinary activity and common lookalikes, not only staged positive events.
Time to alert
Record the delay from a qualifying event to its appearance on the responder screen.
Camera recovery
Verify offline, obstruction, orientation and reconnect behaviour where configured.
Operator usability
Confirm that personnel can inspect, assign and resolve events without ambiguity.
Define a controlled Martion pilot.
Share the agency, intended sites, responder categories and camera environment for an initial operational briefing.