VISYLIX

Visylix supports thousands of concurrent streams per node, and multi-cluster federation is designed toward very large, city-scale estates. Exact capacity depends on your resolution, codec, frame rate and how many analytics you run per stream, so we validate it on your own hardware during the proof of concept. ONVIF auto-discovery and RTSP auto-detection across 35+ URL patterns automatically find and configure cameras from 15+ brands including OEMs including Hikvision, Dahua, Uniview, Axis, Bosch, Hanwha, and more.

Visylix includes 22 built-in AI analytics: (1) Face Recognition with large-scale face database and fast matching, (2) License Plate Recognition (ANPR/ALPR) reading global formats with an India-optimized pipeline, (3) Object Detection with 80+ categories, (4) Person Tracking with per-camera tracking plus v1 cross-camera appearance journeys, (5) Crowd Detection, (6) Safety Gear/PPE Detection, (7) Heat Map Analytics, (8) Motion Detection with adaptive baselines, (9) Pose Estimation with fall detection, (10) Unique Person Counting, (11) Intrusion Detection, (12) Line Crossing Detection, (13) Camera Tampering Detection, (14) Abandoned Object Detection, (15) Tailgating Detection, (16) Queue and Wait-Time Analytics, (17) Parking Occupancy, (18) Speed Estimation, (19) Fire and Smoke Detection, (20) Weapon Detection, (21) Audio Event Detection, and (22) Anomaly Detection. The parts that adapt are specific and bounded: per-camera scene baselines for motion, audio, crowd and tamper calibrate to your environment, anomaly detection trains a baseline per camera, and operators receive threshold retune suggestions they approve before anything changes. Detection accuracy depends on camera angle, lighting and scene, so we validate each analytic on your own footage during the proof of concept.

Yes. Visylix supports on-premise, private cloud, hybrid, and fully air-gapped deployment. The entire system operates without internet after initial installation. Installation takes under 5 minutes with a single command that auto-detects hardware and tunes database pools, worker counts, and connection limits. Online, offline, and Windows installers are available.

Visylix supports 13+ streaming protocols natively in a first-party engine, rather than wrapping FFmpeg or an open-source streaming server. Ingest protocols: RTSP (with Digest Auth), RTMP/RTMPS, SRT, ONVIF (Profile S/T/G), NDI, MPEG-TS, RIST, and GB/T 28181. Output protocols: WebRTC (WHEP/WHIP) with sub-second live view, HLS and LL-HLS with adaptive bitrate, RTSP, HTTP-FLV, HTTP-TS, and DASH. Includes automatic TURN relay, TCP-only TURN for 4K cameras, and per-connection adaptive quality.

Yes. Visylix includes dynamic privacy masks enforced at the server level before video reaches any viewer. It provides on-premise data sovereignty, air-gap compatibility, configurable retention policies, RBAC with 30+ permissions across 4 roles, 2FA, evidence locks with SHA-256 integrity verification and chain of custody, and a comprehensive audit trail logging every action with timestamp, IP, and user agent.

Radha is a conversational AI assistant built specifically for video management, and the only VMS copilot we know of that is conversational, action-taking and fully on-premise at once. It runs 100% on-premise with 77 integrated tools across 8 categories: camera management, recording control, snapshot operations, AI analytics, system monitoring, automation, video playback, and navigation. It supports multi-turn conversations, role-based access, prompt injection protection, per-user rate limiting, and timezone-aware time references.

Yes. Visylix provides NVR forensic access to recordings stored on external NVR devices from 15+ brands including OEMs including Hikvision, Dahua, Uniview, Axis, Bosch, Hanwha, Vivotek, Reolink, TP-Link VIGI, Tiandy, Milesight, Pelco, Avigilon, and any ONVIF Profile G device. Features include time-range search, multi-channel forensic grid, variable-speed playback up to 8x, and recording download with SHA-256 verification.

Visylix offers 22 native AI analytics (face recognition, ANPR, crowd detection, person tracking, pose estimation, PPE detection, and more) that Milestone XProtect requires third-party plugins for. Visylix handles thousands of concurrent streams per node with an optimized architecture, sub-second live view over WebRTC, and flexible cloud/on-premise/edge/hybrid deployment. Visylix has transparent per-stream pricing without per-server licensing.

Visylix integrates 22 AI analytics directly into the VMS without requiring Genetec's separate Analytics modules. Visylix offers edge AI deployment on supported edge compute devices, no per-camera pricing, and thousands of concurrent streams per node, making it a cost-effective alternative to Genetec Security Center.

While Wowza is a general-purpose streaming server, Visylix combines sub-second live view over WebRTC with 22 integrated AI analytics for surveillance. Visylix supports RTSP, RTMP, HLS, WebRTC, SRT, and ONVIF with real-time AI processing, making it an all-in-one streaming + analytics + VMS platform that replaces Wowza plus separate analytics tools.

Visylix offers enterprise-grade AI video analytics integrated directly into the streaming pipeline, with 22 AI analytics including face recognition, ANPR, crowd detection, and more. Ant Media Server is a general-purpose WebRTC server without built-in AI analytics. Visylix handles thousands of concurrent streams per node with GPU inference and supports cloud/on-premise/edge/hybrid deployment.

Unlike Verkada's cloud-only, proprietary-hardware model, Visylix supports cloud, on-premise, edge, and hybrid deployment with any ONVIF camera. Visylix offers 22 AI analytics, thousands of concurrent streams per node, no mandatory cloud subscription fees, and full data sovereignty, making it ideal for organizations with compliance requirements that Verkada cannot meet.

Visylix includes 22 built-in AI analytics that Nx Witness (Network Optix) requires third-party plugins for. Visylix's optimized architecture scales with thousands of concurrent streams per node, hardware-accelerated AI inference, sub-second live view over WebRTC, and enterprise-grade support.

Visylix provides real-time AI analytics during live streaming, while BriefCam focuses on post-event video synopsis. Visylix is a complete VMS + AI analytics platform with 22 AI analytics, thousands of concurrent streams per node, and integrated video management, eliminating the need for a separate VMS alongside BriefCam.

Visylix supports full on-premise and hybrid deployment that Eagle Eye Networks and Rhombus's cloud-focused models cannot. Visylix works with any ONVIF camera (not proprietary hardware), includes 22 built-in AI analytics, and has no per-camera cloud fees. It handles thousands of concurrent streams per node with enterprise-grade scalability.

While Flussonic and Nimble Streamer are general-purpose streaming servers, Visylix integrates 22 AI analytics directly into the streaming pipeline for real-time intelligent video processing. Visylix is a complete enterprise VMS platform with AI analytics, recording, event management, and sub-second live view over WebRTC, not just a streaming server.

Visylix handles thousands of concurrent streams per node without per-camera pricing. Visylix offers 22 production-ready AI analytics with hardware-accelerated inference, full on-premise/edge/hybrid deployment options, and a complete VMS platform including streaming, recording, and integration APIs.

Yes. Visylix is an enterprise-grade platform that scales far beyond Blue Iris's single-server limitations. Visylix handles thousands of concurrent streams per node, offers 22 professional AI analytics (face recognition, ANPR, crowd detection, intrusion detection), multi-site management, hardware-accelerated inference, enterprise support, and cloud/on-premise/edge/hybrid deployment.

Visylix evidence management provides court-admissible exports with SHA-256 integrity hashing, cryptographic chain of custody, and tamper-evident locks. Every evidence action (view, export, share) is written to an immutable audit trail with timestamp, user, IP, and action metadata. Advanced case sharing supports multi-recipient watermarking, time-boxed 24-hour links (basic tier) or role-scoped permanent links (Pro+), and GDPR-compliant right-to-delete workflows.

Yes. Enterprise deployments include HA with primary-standby replication and operator-initiated standby cutover, keeping video ingestion, AI inference, and Radha AI Copilot online during hardware or network failures. Operator-initiated cutover switches traffic to the standby with a documented runbook: RTO under 60 seconds and RPO under 5 seconds, with configurable health checks and DNS-based or L4 load balancer cutover. Fully automatic failover without operator action is on the roadmap, not shipping today.

Visylix supports unlimited polygonal detection zones, directional virtual tripwires for line-crossing counting, and per-zone AI rules (which models run, which thresholds apply, which schedule activates). Zones are drawn directly on the camera preview with per-zone sensitivity, per-zone alerts, and per-zone recording triggers. Tripwires count entries and exits separately with direction awareness.

Visylix delivers alerts across five channels simultaneously: push notifications (via Visylix Cloud), SMS (Twilio integration), email with inline thumbnails, webhooks for custom endpoints, and MQTT for IoT/SCADA systems. Per-user subscription preferences, quiet-hours scheduling, severity routing, and acknowledgment tracking ensure alerts reach the right people without noise.

Enterprise tiers include automatic archive tiering to S3-compatible storage (AWS S3, Cloudflare R2, Wasabi, Backblaze B2). Recordings age from hot local storage to warm and cold archive tiers based on configurable policies (age, camera, event type). Archived footage remains instantly searchable via metadata and can be pulled back on-demand for investigation.

Yes. Visylix ingests access-control events from HID (Origo/Mercury), Lenel S2 / OnGuard, AMAG Symmetry, Brivo, and Genea/Openpath via signature-verified webhooks, parsing each vendor's native event format. Every door event is automatically bookmarked against the mapped cameras, giving a unified badge-and-video timeline for investigations.

Yes. Visylix streams alerts, detections, and audit events to SIEM platforms (Splunk, Elastic, Datadog, QRadar) via syslog, HTTPS webhook, or direct HEC. For PSIM, Visylix exposes outbound bridges to Bosch BIS, Genetec Security Center, and any PSIM that accepts CAP, JSON, or syslog events, centralizing video into command-and-control dashboards.

Yes. Visylix performs real-time fisheye dewarping for 360° and 180° panoramic cameras across four output modes: rectilinear (corrected flat view), panorama (single horizontal strip), quad (four-region split), and dual-panorama. Dewarping runs server-side with GPU acceleration so thin clients like web browsers and mobile receive corrected video without local CPU cost.

Yes. Enterprise tiers include a video wall layout engine supporting up to 6×6 grids per monitor, per-cell camera bindings, automatic rotation, event-driven focus-switching, and multi-monitor orchestration for NOC and command-center environments. Layouts are shareable, profile-scoped, and push-synchronized across operator workstations.

Not yet. Native iOS and Android apps are in development and are not available today. The Visylix web interface runs in mobile browsers in the meantime, with live view over WebRTC. The planned app covers live view, event notifications, case sharing, talkback audio, evidence export and biometric login; see the Visylix roadmap for status.

Yes. Pro tier and above include push notifications delivered via Visylix Cloud (with opt-in per-user device registration) and WebAuthn / Passkeys (FIDO2) login, eliminating password reuse and phishing risk. Admins can enforce MFA (TOTP + recovery codes) at the Starter tier and passkey-only login for privileged roles.

Visylix privacy redaction applies server-side before video reaches any viewer. It supports face redaction, license-plate redaction, and full privacy redaction (all content types including bodies and license plates) on Enterprise. Redaction is applied to live viewing, exports, and case sharing, with configurable exemption rules for authorized RBAC roles (e.g., law enforcement can view un-redacted footage with audit logging).

Scale and Enterprise tiers include SSO via SAML 2.0 and OIDC, compatible with Okta, Azure AD / Entra ID, Google Workspace, Auth0, OneLogin, PingIdentity, and any standards-compliant IdP. Multi-site federation allows a single operator login to traverse unlimited Visylix deployments across geographies with centralized RBAC and aggregated dashboards.

Yes. Pro tier and above include official Python and TypeScript SDKs covering the full 200+ REST API. Both SDKs support live stream control, recording management, AI detection queries, camera provisioning, RBAC operations, and webhook subscriptions with fully-typed request/response models.

Yes. The License Plate Recognition (ANPR/ALPR) model does more than read plates. It also classifies vehicle type (sedan, SUV, truck, motorcycle, bus), color, and general direction as a built-in capability, so every vehicle detection carries both a plate string and a rich appearance profile for downstream search and filtering.

Yes. The engine includes adaptive backpressure (dropping non-reference frames while preserving keyframes for slow subscribers) and a circuit breaker that monitors CPU load in real time, gracefully degrading service at 70%, 85%, and 95% CPU thresholds rather than crashing. New viewers receive the cached Group of Pictures instantly, eliminating the "waiting for keyframe" delay typical of other streaming systems.

Yes. Visylix automatically transcodes H.265 (HEVC) to H.264 in real time for browsers and clients that lack HEVC hardware support. Audio transcoding (G.711, AAC → Opus) happens server-side for WebRTC delivery with proper sample-rate conversion and channel remapping, so operators never have to think about codec compatibility.

Yes, and it is designed to get quieter over time. About 10 of the 22 models adapt to your site: face recognition refines each identity's template from confirmed matches with anti-poisoning guards; fire, weapon, PPE, and object detectors calibrate a per-camera false-alarm threshold from operator dismissals, which by design can only make a safety detector more conservative, never less, with a drift-freeze guard that reverts a runaway camera to baseline; and motion, audio, crowd, camera-tampering, and anomaly each learn their camera's own normal. All adaptation is bounded, reversible, operator-supervised, and runs on-premise. An opt-in, human-approved fine-tune queue can improve the detector models with eval-before-swap, so a new model only ships if it beats the old one. Nothing trains automatically and nothing leaves the premises.

Yes, in a version 1 form. Cross-camera appearance search and person journeys ship today: pick a person from any detection and Visylix finds appearance matches on the site's other cameras within a time window and assembles a chronological path with travel times between cameras. Because it is appearance-based (visual similarity), it returns candidates for an operator to review rather than asserting identity automatically. A purpose-trained re-identification model for higher precision is in our training pipeline. For vehicles, plate journeys give exact cross-camera tracking by plate string. Both are available in plain language through Radha and run entirely on-premise.

Two ways. On its own, Visylix counts how many people cross a door, turnstile, or mantrap and alerts on group crossings using directional, entry-side logic, with no extra hardware. When integrated with an access-control system (HID, LenelS2, AMAG, Brivo, Genea, and Bosch are supported), it correlates how many people crossed the door with how many badges were presented and alerts only when more people entered than credentials, the true piggyback signal. The access-control correlation is opt-in.

Yes. Queue analytics report the live service rate and a predicted wait time for someone joining the queue now, computed from how fast the queue is actually clearing, alongside the current queue length. The prediction reflects real conditions at that counter and updates continuously as the line speeds up or slows down.

No. Visylix runs fully on-premise. All AI inference, model adaptation, cross-camera journeys, and analytics happen on your own hardware, and it operates in fully air-gapped deployments after installation. No frames, detections, faces, plates, or learned baselines are sent to any external service. Optional push notifications can be relayed via Visylix Cloud only if you explicitly enable them.