Product Overview
LILIN EDGEFIRE is an edge-based AI video analytics software license for compatible LILIN IP surveillance cameras. It provides real-time visual detection of fire, smoke, and smog directly on the camera, helping security teams identify early-stage hazards without relying on a separate high-performance analytics server.
Key Configuration
Model: LILIN EDGEFIRE; product type: front-end Edge AI video analytics software license; detection targets: fire, smoke, and smog; trigger condition: alarm when fire or smoke occupies more than 10% of the field of view; maximum tracked objects: up to 24 objects per frame; processing speed: up to 6 recognitions per second per channel; detection zones: up to 4 custom zones; heartbeat checking: notification support every 1–30 minutes.
Key Features
• Detects active flames, smoke, and smog for early visual fire-warning applications.
• Runs AI processing directly on compatible camera hardware to reduce server dependency.
• Supports up to 4 configurable detection zones for focused hazard monitoring.
• Tracks up to 24 objects per frame with up to 6 recognitions per second per channel.
• Heartbeat checking helps confirm that camera AI operation remains active.
Technical Context
EDGEFIRE is part of LILIN’s Edge Aida AI ecosystem and is dedicated to visual fire, smoke, and smog detection. It can support rules such as prohibited zone monitoring, dwell-based hazard accumulation, compound detection logic, camera tampering alerts, and optional result overlays for live monitoring.
Installation Notes
Install EDGEFIRE on a compatible LILIN AI camera and configure the detection zones around areas where smoke or flame visibility is critical. Test the system under expected lighting, camera angle, and field-of-view conditions, and set heartbeat notification intervals according to the required fire-surveillance uptime policy.
| General | |
| Model | EDGEFIRE |
| Recognized objects | Fire and smoke |
| Object Size Limitation | Support detection when size of smoke or fire is more than 10 % of FOV displayed on the screen |
| Results Display | Support conditional display results according to specified objects |
| Warranty | 1 Year |
| Behaviors Detection | |
| Behaviors Detection | Prohibited Zone : Support triggering an event when specified object(s) enters restricted area Zone with Dwell /Parking Violation : Support triggering an event when specified object(s) enters restricted area and stays for a predefined duration of time Tampering Detection : Support triggering an event when the camera is blocked, redirected, or shaken All Object Detection (AND) : Support triggering an event when all of the specified objects show up within the set detection zone at the same time (based on AND rule) Lack of any Object Detection (NAND) : Support triggering an event when any of the specified objects is not present within the set detection zone (based on NAND rule) |
| Recognition | |
| Recognition Speed | Support object recognition speed up to 12 frames per second at 4K, 14 frames per second at 1080P, and 16 frames per second at 720P |
| Number of Detectable Objects | Can detect up to 24 objects per frame |
| Video resolution | Support up to 4K(3840 x 2160) and 2MP(1920 x 1080) image recognition resolution |
| Minimum Recognition Resolution | Object more than 10 % of FOV ，smoke opacity is more than 80% |
| Objects Filtering | |
| Objects Filtering | Object Type :Support recognizing specified object, instead of processing all types of objects Object Size : Support 0~100% adjustable size of object (compared to frame) for classification Support 0~100% adjustable size of object (compared to frame) for behavior detection Object Confidence Level : Support 0~100% adjustable confidence level of object |
| Events | |
| Event Trigger | Support setting trigger by specific behaviors of specified objects Support setting 8 Cold Zones to minimize false alarms and improve the accuracy |
| Detection Time | After an event is triggered by specified object, the next detection will be carried out according to the set detection time (2~30 seconds, 10 minutes, 20 minutes, 30 minutes (AI Missing Object Detection and Tampering Detection will not trigger the next detection) |
| Heartbeat Checking | Support checking the camera AI operation status and pushing notifications every 1~30 minutes |
| Detection Zone / Line | |
| Detection Zone / Line | Support setting 4 detection zones (about the size of camera FoV) or lines which are influenced by independent filter condition (object size, confidence, dwell, classification and behavior) and to trigger independent events (output) |
| Polygon Area | Support detection zone which is a polygon that can be freely dragged and that the shape can be modified by 6 draggable anchor points Support detection line which is a line attached to a polygon that the shape can be modified by 4 draggable anchor points. Also, direction of the line can be set by dragging the 4 anchor points or clicking the “Enable direction button” |
| Pixels Calculation | Support displaying the pixel count of selected object(s) on the web UI |
| Integration | |
| Alarm Triggering | Digital Output : Support triggering digital output of camera Smart Event : Support triggering up to 4 virtual inputs of camera HTTP POST : Support displaying an alarm on LILIN IP LED Panel Support displaying an alarm on specific channel of LILIN Navigator VMS Support triggering digital outputs for LILIN I/O Box Support triggering digital outputs for LILIN NVR Support triggering HTTP POST to integrate with third-party devices Support triggering a PTZ camera to go to a preset |
| Recording System Integration | Support event search to playback or snapshot on LILIN Navigator VMS and NVR |
| API Integration | Support API integration with third-party VMS and provide camera SDK files |
| Interface Language | Support Traditional Chinese, English, Japanese and Italian |
| Camera supported | LILIN Z7/P7 series cameras |
| Camera supported | Z7/S7 |