Our website use cookies to improve and personalize your experience and to display advertisements(if any). Our website may also include cookies from third parties like Google Adsense, Google Analytics, Youtube. By using the website, you consent to the use of cookies. We have updated our Privacy Policy. Please click on the button to check our Privacy Policy.

How to Test Wildfire Detection System

We worked with our partners to conduct fire tests in different markets over the years – Hong Kong, Indonesia, Thailand, Vietnam, Canada, Mexico, Brazil, Uruguay, Portugal, Latvia, Greece, etc. We did fire tests for several purposes: product evaluations, UAT (user acceptance test), annual product maintenance, and AI model training.

We will do fire tests in the coming months. It is a good time to share some of our fire test best practices*. We will focus on image analytics-based wildfire detections.

Thermal detection of burning vegetation

Our thermal detection identifies heat sources in thermal images captured by LWIR (long-wavelength infrared) thermal sensor. The temperature of burning vegetation is between 300 and 800 degrees Celsius. The burning plants will emit more intense infrared radiation than the surrounding environment.

Thermal detection detects early-stage forest fires after the initial ignitions. It is fast and accurate. We can detect a burning vegetation surface as small as 2-meter square size at a 5km distance. Infrared radiation, like all other radiations, travels in a straight line. Hence thermal detection requires LOS (line of sight) between the burning object and the thermal sensor.

With all the safety measures in place, if the detection distance is 2km or less, we suggest the following field test setup: wood logs with dimensions 1 m x 1 m x 1 m. The wood logs should be around 10 cm to 15 cm thick in diameter and should be arranged and stacked as shown below:

Figure 1: fire test setup for thermal detection of forest fire

This method of stacking the wood logs has the advantage of ensuring that the burning lasts for at least 30 minutes. Smaller wood branches and gasoline are added to help the ignition. The fire may take 5 minutes to 15 minutes to reach its maximum intensity.

The following YouTube video shows a fire test we conducted in Greece.

Visual detection of wildfire flame and smoke

Our AI wildfire detection uses Machine Learning object detection to detect forest fires. We used about half a million visual images of wildfires to train our AI models. They are small flame or smoke photos of distance surface fires. The smoke rise above tree canopies is a important visual signal of distant forest fire.

Figure 2: smoke of distant fire

With all safety measures implemented, we suggest the following field test setup: wood logs with dimensions 1 m x 1 m x 1 m. The wood logs should be around 10 cm to 15 cm thick in diameter and should be arranged and stacked. When the wood logs are burned,  completely charred and visible flames extinguished, add damp green leaves and grass on top of the charred wood left over. White smoke will be generated as shown below:

Figure 3: fire test setup for AI visual detection of forest fire

Water should be sprayed on the greens from time to time to prevent them from drying and starting to burn. The smoke column can be generated for at least 15 minutes.

Figure 4: a recent fire test we conducted to evaluate our AI wildfire detection

LookOut “Test Detect” tool

Field tests are good to evaluate, demonstrate, and train our wildfire detection system. However, it takes time to arrange it. And we cannot do it during fire season. What if I need a wildfire detection system to protect my community next week? How can I test the technology?

Our LookOut wildfire detection service provides cloud-based AI wildfire detection. It does not matter to LookOut if the visual image is sent from an AXIS Q6215-LE in Piriapolis or a computer in Montevideo. LookOut applies the same Machine Learning object detection on both images. You can use the LookOut built-in Test Detect feature to send an image from your computer to LookOut and test it, anytime and anywhere. Test Detect is a very handy tool for our customers to test and try our technologies.

Figure 5: LookOut built-in Test Detect tool

Detection of Candle Fire


We cannot detect it. We do not build our AI model to do it.

Figure 6: A peaceful candle fire. We need more peace.

*Please feel free to contact us at info@roboticscats.com if you want a copy of our Standard Fire Test Procedures.

Related Posts