Technology Overview

Fire Detection is a video surveillance technology designed to automatically identify visual or thermal indications of a potential fire and generate real-time events for faster awareness and response.

ACTi video surveillance solutions can provide two complementary approaches to Fire Detection:

AI-Based Fire Detection uses regular surveillance cameras and AI video analytics to recognize the visual appearance and dynamic behavior of flames.

Thermal Fire Detection uses thermal cameras to monitor temperatures and generate events when a monitored object or area reaches or exceeds a configured temperature threshold.

Customers can select either approach according to the requirements of the monitored environment or deploy both technologies together.

The two methods observe different characteristics of a developing fire:

AI camera → “Are visible flames present?”

Thermal camera → “Has the monitored area reached an abnormal or critical temperature?”

Because these detection methods are based on different physical evidence, they can complement one another when combined.

Two Approaches to Fire Detection AI-Based Visual Fire Detection

AI-based Fire Detection uses video from a regular surveillance camera.

The AI continuously analyzes consecutive video frames for visual characteristics associated with flames, including:

Flame-like colors and brightness Irregular flame shapes Rapidly changing boundaries Flickering Upward or turbulent movement Repeated shape deformation Growth or spread over time Persistent fire-like visual patterns

Rather than relying on a single characteristic, the AI evaluates a combination of appearance and temporal behavior to determine whether the observed visual condition resembles fire.

Thermal Fire Detection

Thermal Fire Detection uses a thermal camera capable of monitoring temperature information within the scene.

Instead of recognizing the visual appearance of flames, the thermal camera monitors heat emitted by surfaces and objects.

Temperature thresholds can be configured for relevant monitoring areas or targets.

When the measured temperature reaches or exceeds the configured threshold, the system can generate a Fire Detection event.

Conceptually:

Temperature below threshold → normal condition

Temperature reaches/exceeds threshold → thermal event generated

This makes thermal cameras particularly useful for identifying abnormal heating conditions that may occur before visible flames become obvious.

How AI-Based Fire Detection Works

The AI continuously processes video frames from a conventional surveillance camera.

Fire is visually different from rigid objects such as people or vehicles because its shape changes continuously.

Flames may:

Expand and contract Flicker Split and merge Rise Bend with airflow Change brightness Spread across combustible material Rapidly change their boundaries

The AI evaluates these visual characteristics across multiple frames.

A simplified detection sequence can be represented as:

Normal scene → fire-like region appears → flame characteristics analyzed → temporal behavior confirmed → Fire Detection event

Using temporal analysis helps distinguish genuine flames from ordinary objects that may resemble fire in a single image.

Flame Appearance and Shape Analysis

Color and brightness can provide useful evidence of fire.

Depending on combustion conditions, flames may contain combinations of yellow, orange, red, white, or blue regions.

However, color alone cannot reliably identify fire.

Ordinary objects such as lamps, vehicle lights, signs, clothing, reflections, or illuminated displays may contain similar colors.

The AI therefore evaluates color together with the characteristic irregular shape and changing boundaries of flames.

A solid orange object may maintain approximately the same shape over time, while a flame continuously changes.

This dynamic behavior provides additional information for AI classification.

Flickering and Motion Analysis

Flickering is another characteristic visual behavior associated with flames.

The brightness, shape, and position of a flame can change rapidly as combustion and airflow interact.

The AI can analyze these variations across consecutive frames and determine whether they correspond to learned flame behavior.

Motion can provide further evidence. Flames frequently exhibit upward or turbulent movement, although wind and surrounding conditions can alter their direction.

The AI therefore evaluates movement together with appearance, shape, and temporal behavior rather than relying on a simple motion rule.

Thermal Imaging and Temperature Measurement

Thermal cameras provide a fundamentally different method of Fire Detection.

Instead of analyzing visible light, a thermal camera detects infrared radiation emitted by surfaces and objects and converts this information into a thermal image.

Where supported by the thermal camera, temperature information can be evaluated against configured limits.

For example, an area containing stored material may normally remain below a particular temperature.

If part of the monitored material begins heating abnormally and exceeds the configured threshold, the thermal system can generate an event.

This approach does not require visible flames to be present.

Temperature Thresholds

Temperature thresholds allow thermal Fire Detection to be adapted to different applications.

For example, one monitored object may normally operate at relatively high temperatures, while another storage area should remain close to ambient temperature.

The appropriate threshold therefore depends on:

Type of material or equipment Normal operating temperature Ambient environmental conditions Required warning level Monitoring objective

The threshold should be configured according to the expected thermal characteristics of the specific target rather than applying one universal temperature value to every scene.

Hotspot Detection

Some potential fire conditions begin with localized heating.

Examples can include overheating machinery, electrical equipment, batteries, stored materials, waste, or recycling materials.

A thermal camera can identify areas that are significantly hotter than their surroundings.

Potential applications include:

Waste and recycling facilities Material stockpiles Battery storage areas Electrical installations Industrial equipment Warehouses Outdoor storage yards Production facilities

Detecting an abnormal hotspot can allow personnel to investigate a developing condition before it progresses into a larger incident.

Visible Flames and Abnormal Heat Are Different Indicators

AI-based and thermal Fire Detection should be understood as complementary technologies because they observe different evidence.

AI-based Fire Detection detects the visible characteristics of flames.

Thermal Fire Detection detects temperature conditions.

A high-temperature condition does not necessarily mean that visible flames are present.

For example, machinery or electrical equipment may overheat without igniting.

Similarly, visible flames may be present while the relevant thermal target is partially obscured or outside a configured temperature-monitoring area.

This distinction is important when selecting the appropriate technology.

Choosing AI-Based Fire Detection

AI-based Fire Detection is particularly suitable when:

Visible flames are expected to be observable Regular surveillance cameras already cover the area Wide-area visual monitoring is required Immediate visual verification is important Outdoor or semi-outdoor monitoring is required Potential fire sources can be clearly viewed by the camera

Because the technology uses conventional surveillance video, the same camera can provide security monitoring, recording, and AI Fire Detection.

Choosing Thermal Fire Detection

Thermal Fire Detection is particularly suitable when:

Abnormal temperature may develop before visible flames Temperature thresholds can be meaningfully defined Equipment or materials are vulnerable to overheating Hotspot detection is important Monitoring is required in darkness Visible-light conditions are difficult or inconsistent Preventive thermal monitoring is required

Thermal imaging can be especially useful where the objective is to identify a potentially dangerous temperature condition before obvious combustion occurs.

Combining AI and Thermal Fire Detection

For higher-value or higher-risk applications, customers can deploy both technologies.

A combined solution provides two different sources of information:

Regular camera → AI detects visible flame behavior

Thermal camera → temperature threshold identifies abnormal heat

These analytics can operate independently while contributing events to the same surveillance and monitoring environment.

For example:

Abnormal temperature detected → thermal alert generated → personnel investigate

or:

Visible flames detected → AI Fire Detection alert generated → personnel investigate

In some incidents, both technologies may generate corresponding events.

The combined approach can provide additional situational information because one technology may identify conditions that are less apparent to the other.

Complementary Verification

When both camera types cover the same general area, operators can use visible and thermal imagery together.

The regular camera can provide visual context such as:

Visible flames Smoke Nearby people Equipment Materials Scene activity

The thermal camera can provide information about:

Hotspots Elevated surface temperatures Location of abnormal heating Thermal development over time

Together, these views can help operators understand the nature and location of a potential incident more quickly.

Outdoor and Semi-Outdoor Environments

Video-based Fire Detection can provide particular value in outdoor and semi-outdoor environments.

Potential applications include:

Outdoor storage yards Recycling facilities Waste-handling areas Construction sites Logistics yards Industrial compounds Agricultural facilities Material stockpiles

AI cameras can monitor these areas for visible flames across a wide field of view.

Thermal cameras can monitor selected materials, equipment, or areas for abnormal temperatures.

Customers can choose the technology most appropriate to the risk or combine both approaches.

Large Indoor Spaces

Large-volume buildings can also benefit from camera-based Fire Detection.

Potential environments include:

Warehouses Aircraft hangars Distribution centers Production halls Large storage facilities

AI cameras may provide direct visual observation of developing flames, while thermal cameras can monitor high-risk equipment or stored materials for abnormal heating.

This allows the surveillance system to provide additional fire awareness across areas that may be difficult to monitor continuously by personnel.

Day and Night Operation

AI-based Fire Detection depends on the visible image produced by the surveillance camera.

Flames themselves emit visible light and can remain clearly visible at night, although artificial lighting, headlights, reflections, and other bright objects can create additional analytical challenges.

Thermal imaging does not depend on conventional visible illumination.

A thermal camera can monitor temperature differences in daylight, darkness, and many low-light conditions.

For locations requiring continuous monitoring through significant lighting changes, thermal imaging can therefore provide an important complementary capability.

Detection Zones

Both AI and thermal Fire Detection can focus on relevant monitoring areas.

For AI-based detection, a configured zone can restrict flame analytics to parts of the camera view where uncontrolled fire is a concern.

For thermal monitoring, temperature rules can be applied to relevant targets or regions according to the capabilities of the thermal camera and analytics system.

Proper zone configuration can help exclude known sources of irrelevant activity.

For example, a regular camera monitoring an industrial facility might exclude an area where controlled welding routinely occurs.

Similarly, thermal monitoring can focus on specific machinery or material-storage areas rather than unrelated surfaces.

Camera Placement for AI Fire Detection

For visual Fire Detection, the camera should have a clear view of potential fire-source areas.

Important considerations include:

Clear line of sight Sufficient flame size within the image Appropriate monitoring distance Suitable field of view Adequate image quality Appropriate exposure Minimal obstruction Stable camera mounting

A very wide field of view may cover a large area but make small flames difficult to distinguish.

Camera placement should therefore balance area coverage with sufficient visual detail.

Camera Placement for Thermal Fire Detection

Thermal camera installation should be designed around the thermal target.

Important considerations include:

Clear line of sight to the monitored surface or object Suitable monitoring distance Sufficient target size within the thermal image Expected normal temperature range Ambient temperature variation Surface characteristics Reflective materials Weather exposure Required temperature accuracy

Thermal cameras generally measure infrared radiation from visible surfaces; they should not be assumed to see through walls or other solid obstructions.

Environmental Factors

AI-based Fire Detection can be affected by conditions such as:

Reflections Bright lights Welding Sparks Strong sunlight Poor visibility Fog Heavy rain or snow Camera vibration Low-resolution video Physical obstruction

Thermal Fire Detection has different environmental considerations, including:

Ambient temperature Solar heating Surface emissivity Reflections from certain materials Weather conditions Monitoring distance Thermal resolution Target visibility

The most appropriate technology should therefore be selected according to the characteristics of the deployment environment.

Confidence and Detection Parameters

AI-based Fire Detection is probabilistic.

The AI evaluates how closely observed visual characteristics correspond to flame behavior represented by the detection model.

Sensitivity and confidence parameters can be adjusted to balance detection capability and resistance to false alarms.

Thermal Fire Detection instead relies on configured thermal conditions such as temperature thresholds.

These thresholds should be selected according to the normal operating characteristics and risk level of the monitored target.

In both cases, site-specific configuration and testing are important.

Event Generation and Integration

When either detection technology satisfies its configured criteria, the system can generate a Fire Detection event.

Events can be integrated into the broader ACTi video surveillance workflow and may be used to:

Display an immediate alert to an operator Highlight the corresponding camera Trigger or bookmark video recording Generate an alarm or notification Create an event record for later search Provide visible or thermal video for verification Initiate other configured safety or emergency workflows

If both AI and thermal detection are deployed, events from both technologies can contribute to the overall monitoring workflow.

Real-Time Monitoring

Both technologies can provide continuous automated monitoring.

For AI-based detection:

Video analysis → flame-like region detected → temporal characteristics evaluated → confidence criteria satisfied → event generated

For thermal detection:

Temperature monitored → threshold reached or exceeded → thermal event generated

This allows the surveillance system to continuously monitor relevant areas without requiring operators to watch every camera at all times.

Technical Benefits

The availability of two Fire Detection technologies provides flexibility for different applications.

Key benefits include:

AI-based visible flame recognition Thermal temperature-threshold monitoring Hotspot and abnormal-heat awareness Customer choice between AI and thermal technologies Ability to combine both technologies Wide-area visual monitoring Day and night thermal monitoring Configurable monitoring zones Configurable temperature thresholds Real-time event generation Immediate visual or thermal verification Integration with alarms and notifications Event-associated video recording and search Support for indoor, outdoor, and semi-outdoor environments Performance Considerations

Both Fire Detection methods have different strengths and limitations.

AI-based Fire Detection requires sufficient visible evidence of flames. Detection may be difficult when flames are hidden, extremely small, distant, obscured by dense smoke, or outside the camera's field of view.

Visually similar phenomena such as welding, reflections, flashing lights, or controlled flames can also create challenging conditions.

Thermal Fire Detection requires a suitable view of the monitored thermal target and correctly configured temperature criteria. Environmental conditions, target surface properties, monitoring distance, thermal resolution, and reflections can influence temperature measurements.

Neither technology should therefore be expected to detect every possible fire condition under all circumstances.

Using both technologies can provide complementary information, but it does not eliminate the need for appropriate system design, configuration, and testing.

Relationship to Conventional Fire Detection Systems

AI-based and thermal Fire Detection can provide valuable additional early-warning, monitoring, and verification capabilities, particularly in large, outdoor, semi-outdoor, industrial, and other challenging environments.

However, surveillance-based AI and thermal analytics should not automatically be considered replacements for certified fire detection, alarm, suppression, or life-safety systems required by applicable regulations, codes, or standards.

Where certified fire protection is required, the appropriate approved systems should remain in place.

Video and thermal Fire Detection can complement these systems by providing automated wide-area monitoring and immediate visual information about suspected incidents.

Selecting the Right Approach

The appropriate Fire Detection architecture depends on the application.

AI-based Fire Detection is a strong choice when visible flames can be clearly observed and regular surveillance cameras provide suitable coverage.

Thermal Fire Detection is a strong choice when abnormal temperature or hotspots provide an important early indication of risk.

Combined AI + Thermal Fire Detection is suitable when both visible flame recognition and temperature monitoring provide valuable complementary information.

This flexibility allows customers to select the solution according to the monitored environment, operational requirements, available camera infrastructure, and fire risks.

Conclusion

ACTi Fire Detection can use AI-based visual analytics, thermal temperature monitoring, or a combination of both technologies to provide automated awareness of potential fire conditions.

Regular surveillance cameras equipped with AI analytics recognize the visual appearance and dynamic behavior of flames, while thermal cameras identify abnormal heat conditions using configurable temperature thresholds.

Each approach provides different information and can be deployed independently. When appropriate, combining the two provides complementary visual and thermal perspectives of the monitored environment.

With suitable camera selection, positioning, detection zones, temperature thresholds, analytics configuration, and site-specific testing, customers can choose the Fire Detection architecture that best matches their application while integrating detected events with the broader ACTi video surveillance workflow.