What's New at FireVision

Detection History Filtering: Completely overhauled the detection history on the dashboard. You can now seamlessly filter events by type (Fire & Smoke, Human, Vehicle, Animal) using scrollable tabs on desktop or the dropdown menu on mobile.

Off-Grid Remote Hubs: Launched new hardware packages for completely off-grid setups, featuring a remote hub bundled with a 4G SIM router, solar panel, and battery system.

ROI Calculator & Case Studies: Added an interactive ROI calculator to help calculate potential savings from false alarm reductions, alongside new detailed case studies focused on logistics and warehouse environments.

Glossary & Governance: Added a dynamic Glossary feature and dedicated governance details to the About page to enhance platform transparency and education.

Camera Naming Flexibility: The backend video processor has been upgraded to gracefully support highly customized, user-defined camera names and complex filename structures without interrupting the video syncing process.

Discoverability & SEO: Deep architectural upgrades including a new dynamic Glossary, IndexNow API integration for instant search engine crawling, and advanced structured data schemas (AEO/GEO) for AI search discoverability.

Missing Clip Synchronization: Deployed a new background synchronization service that automatically detects and recovers any video clips that may have dropped due to temporary network interruptions.

Detection Engine Stability: Implemented graceful shutdowns and enhanced error handling in our core video stream processing engine to ensure uninterrupted monitoring and prevent segmentation faults during routine maintenance.

Official Hardware Store: Launched a dedicated Product Hub! You can now purchase guaranteed-compatible hardware directly from us, including pre-configured VPN Hubs and our recommended Wansview 2K Security Cameras.

Semantic SEO (AEO/GEO): Continuous enhancements for Answer Engine Optimization based on AI research to improve the platform's trusted expert ranking.

Camera Compatibility Guidance: Updated our documentation and help files to clearly outline supported camera ecosystems. We explicitly require RTSP/ONVIF streams and have listed proprietary, cloud-locked brands that are incompatible.

MultiVision Detection Models: The backend now supports secondary detection weights, allowing the AI to simultaneously monitor for humans, vehicles, and animals alongside standard fire and smoke detection. Added interactive demos to the homepage.

B2B Semantic API Endpoints: Introduced Agent.json and OpenAPI schemas to allow autonomous AI agents to introspect our data structures and rate limits safely.

Comprehensive Audit Logging: We completely overhauled our internal logging. Every profile update, camera tweak, and notification change is now perfectly sequenced on the backend for absolute transparency.

Database Unification: Merged disparate detection tables into a single unified `detectionLogs` schema to streamline performance and provide a single source of truth for your history.

Super Admin Server Alerts: Upgraded our Vultr server monitoring processes and implemented dedicated email alerts for high-usage scenarios and ghost container cleanups.

Live Stream Stability: Resolved streaming latency and buffering issues on the Live page. Added a 30fps fallback safety net for the video metadata decoder to prevent recording crashes.

Smart Error Throttling: Implemented throttled logging on live camera feeds to prevent duplicate connection warnings from overwhelming the interface.

FireVision V1.5: A complete visual overhaul of the platform featuring a highly requested, system-wide Light/Dark mode toggle and modernized UI elements.

Precision Camera Scheduling: Camera scheduling has been revamped. You can now assign specific timezones and active hours to individual cameras, giving you precise control over when detection is active.

International Phone Routing: Mobile numbers now use dedicated country-code selectors during setup to prevent international routing issues for SMS and Call alerts.

Plan Allocation Logic: Addressed bugs relating to camera addition limits for users who have registered but not yet selected a subscription plan.

Distributor Dashboards: Introduced dedicated portals allowing regional partners to provide managed support and account updates safely for enterprise clients.

Critical Security Upgrades: Applied critical security patches and upgraded our core web framework to ensure maximum platform security against the latest CVE vulnerabilities.

Pre-detection Recording Buffer: The video processing engine now maintains a 2-second pre-detection rolling buffer, ensuring you capture the very beginning of an event prior to the alarm threshold being met.

Live Chat Integration: Integrated a powerful n8n live chat support widget to provide immediate assistance directly on the platform.

Smart Detection Cooldown: Implemented an intelligent cooldown period for continual detections to prevent alert fatigue while maintaining constant monitoring.

Subscription Visibility: The profile dashboard now displays clear camera allocation limits with seamless upgrade pathways.

Automated Camera Management: Background PM2 updater scripts now automatically sanitize and route new camera name additions straight to the MediaMTX configurations.

Stripe Subscriptions: Fully integrated Stripe payment processing, intent routing, and webhooks to support seamless plan upgrades and automated billing.

Automated Video Deletion: Deployed robust PM2 background cleanup services that automatically purge old .mp4 detection videos from the Vultr server based on user-defined retention limits.

Server Health Monitoring: A new "Server Resource Usage" dashboard allows administrators to monitor the real-time load, memory, and ghost containers of the underlying server infrastructure.

Detection Accuracy: Increased the base confidence threshold to 30% for improved accuracy and significantly fewer false positives in complex lighting.

Offline Camera Alerts: The system now identifies multi-camera support structures and dispatches targeted email alerts if an active camera goes offline or drops its stream.

Multi-Channel Alerts: Integrated Twilio for automated Phone Call and SMS notifications, running alongside our newly implemented Nodemailer service for instant, reliable event dispatching.

Off-grid/No-DDNS Support: Introduced mandatory External IP and Port fields for users lacking DDNS capabilities, vastly expanding the types of networks we can securely monitor.

User Activity Logging: Integrated server-to-server webhook tracking between Clerk and Convex to provide an immutable, transparent log of user logins, logouts, and system changes.

Intelligent Live Streaming: To conserve data and Vultr server processing, live streams now automatically pause after five minutes of inactivity or loss of page focus.

Vultr Docker Stability: Reconfigured Nginx mount points, permissions, and file naming conventions within the Vultr docker container to ensure flawless playback routing.

Automated LiveCam: Introduced real-time HLS playback with automatic MediaMTX config generation from profile data, removing the need for manual file tweaking.

Event-based Video Trimming: Implemented background FFmpeg sub-processes that cleanly trim and segment only the parts of the video containing smoke or fire, discarding the rest to save space.

Mobile-Responsive Architecture: A sweeping UI update introducing hamburger menus, new branding, Google reCAPTCHA forms, and improved layout visibility across devices.

Architecture Separation: Successfully separated the Frontend (Vercel) and Backend (Vultr) applications to run entirely independently.

Initial AI Training: Established the foundational YOLOv5 machine learning infrastructure and defined validation pipelines using dfire.yaml datasets.

Secure User Authentication: Partnered with Clerk to implement frictionless, secure user onboarding, authentication, and session handling.

Cloud Database Integration: Implemented the core Convex database to act as the reactive, real-time backbone linking the UI to our backend AI engines.

Admin Analytics Dashboard: Created the first iteration of the super-admin metrics panel for live oversight.