Your cameras are already installed and recording. The question is who — or what — is actually watching them. WillFly AI Box is an industrial edge computing device that adds intelligent video analysis to the surveillance cameras you already have: it analyzes every feed locally, around the clock, and turns what it sees into alert events your team can verify, assign and close.
One browser-based console covers the complete workflow — live view, per-camera rule configuration, alert handling, evidence export and system health. Four numbers to remember: 7×24 analysis, 4+ access protocols, second-level alerts, and video data that stays on site.
1. How It Works
Connect your existing cameras over RTSP, ONVIF, RTMP or GB28181 — four protocols covering mainstream brands and video platforms.
Select a detection model: flame and smoke, crowd gathering, and more.
Configure rules per camera: detection zones, virtual lines, schedules and alert levels.
Analyze continuously at the edge — events are generated in seconds, not after a round trip to the cloud.
Verify and close: check the snapshot evidence, assign a handler and record the outcome.
All five steps run in one browser console. No separate apps, no cloud account required.
2. Built Around the Cameras You Already Have
The box connects to your current surveillance system instead of replacing it. Live preview supports single and multi-camera layouts, with one-click switching between the raw feed and the analyzed view. Device online status and bound models are visible at a glance, and batch import and export make larger deployments a configuration job rather than a project.
Detection rules are set per camera, not per site: draw zones and virtual lines on the actual image, then set duration, cooldown time, alert level and active schedule for each point. One model can serve many cameras, each with its own rules — so an entrance, a corridor and a work area can each trigger the way they should. Alerts that match reality are the ones your team keeps trusting.
3. AI Capability and Video Access
100+ AI recognition algorithms. The platform covers safety protection, area events, vehicle detection and counting, and more — new scenarios can be configured as your business needs grow.
Up to 32 channels of 1080p cameras — or 16 channels at 2K resolution — analyzed in real time at the edge, around the clock.
Other hardware is configured flexibly per project: tell us how many cameras you run and which events you need detected, and we will confirm the right setup at quotation.
4. Design and Interfaces
The enclosure is aluminum alloy with fanless passive cooling: no moving parts to fail, no dust pulled inside, no noise in control rooms — built for continuous 24/7 operation. Connectivity matches real deployment conditions: Wi-Fi and 4G for sites without wired infrastructure, Ethernet for intranet-first installations, and a rich set of IO ports for site peripherals. Detailed dimensions, mounting options and the full port list depend on the configuration chosen, and are confirmed at quotation.
5. Where It Works
Factories and warehouses — flame and smoke detection. Deploy flame and smoke models at key points, with alert levels and notification rules that surface the frames worth human review. It closes the blind gaps between manual patrols.
Parks and stations — crowd gathering monitoring. People-count thresholds and duration conditions are set per zone, and each area runs its own rules, supporting crowd management and order maintenance.
Existing footage — offline review and sampling. Submit images or video files as offline analysis tasks and review results in the console — useful for routine sampling and for evaluating model performance before live deployment, with no extra hardware.
A scenario library that grows with you. The platform ships with an extension base covering safety protection, area events, vehicle detection and counting — configure new scenarios as your business needs them.
6. Start Small, Then Expand
You do not need to overhaul the whole surveillance system at once. Start with one or two points where the value is clear and the camera view is stable; verify detection against real footage by day and by night; run the full loop from alert to closed event; then expand to more points and scenarios. Model quantity and versions are configured flexibly with each delivery.
7. Why Edge Instead of Cloud
Cloud analysis means uploading raw video around the clock: high bandwidth and storage costs, delay on every event, and footage leaving your premises. Manual monitoring means fatigue and missed moments across dozens of screens. Edge analysis keeps the compute next to the cameras: second-level alerts, only lightweight event data leaving the site, raw video staying on your network, and the whole system running on an intranet with no public cloud dependency.
8. What You Get
WillFly AI Box is coming soon. Standard orders are based on a minimum of 1 set, with a typical lead time of 15 to 30 days depending on configuration. Every unit carries a 12-month warranty with spare parts and remote technical support, and we handle the export documentation for your market. OEM and ODM configurations are available for distributors and system integrators.
Send us your site, your cameras and the events you need detected through our contact page — we will confirm feasibility and configuration within one working day, and notify you at release.
9. Why Source From WillFly
WillFly Drone has built payload and platform systems since 2016: cleaning, seeding and firefighting modules for drones, plus ground robotics, serving customers in 30+ countries with more than 1,000 units delivered. The WillFly AI Box extends the same approach to fixed cameras — it upgrades the equipment you already operate instead of replacing it, and it sits alongside our robot dog platforms for ground inspection and our DJI-compatible drone payloads for aerial work.
10. Questions Buyers Ask
Do I need to replace my existing cameras?
No. The box connects over RTSP, ONVIF, RTMP and GB28181, which cover mainstream camera brands and video platforms.
Does our video go to the cloud?
No. Analysis runs locally at the edge; raw video stays on your network and only event data is shared within the console.
Can we test it on recorded footage first?
Yes. Images and videos can be submitted as offline analysis tasks for batch review, so you can evaluate results before any live deployment.
How do we get a quote?
Send your use case and delivery market through the contact page. We will confirm the configuration and reply within one working day.
