Rosella       Machine Intelligence & Data Mining

Wildfire/Bushfire Detection by AI Machine Learning

Wildfires (aka bushfires in Australia) can occur naturally by lightening strikes. Sometimes caused by arson or human activities. Wildfires can destroy homes, kill people and natural habitats, destroy infrastructure, and pollute air.

Early detection is crucial for containment of wildfires. Drone and satellite images can be used to monitor and detect bushfires. However drone and satellite images can be vast. Impractical to investigate by human eyes. AI Machine Learning can help!

Artificial Intelligence and Machine Learning / Deep Learning

Machine Learning (ML) is a field of Artificial Intelligence (AI). Machine learning is a computer modeling that learns intelligence from data. From drone and satellite images, ML can learn fire detection capability accurately. The following images show fire detection examples. This ML Deep Learning model was developed using OD-CNN (Object Detection - Convolutional Neural Network) of CMSR Machine Learning Studio. Green boxes are detected fire areas. As seen in these examples, CMSR OD-CNN can detect fires with high accuracy.

wildfire/bushfire detection.

Good Training Data is Essential for High Accuracy

Machine learning can detect wildfires accurately provided that similar patterns are included in training datasets. For unseen patterns that are not included in training datasets, results can be random! Thus it is very important to include all predictable patterns in training datasets.

Wildfire Detection Modeling Techniques

There are several ways to model wildfire detection;

  1. Fire vs nofire classification: CNN
  2. Fire probability projection: CNN
  3. Fire, ash cloud and nofire probability projection: M-CNN
  4. Grid cell fire probability projection: OD-CNN
  5. Fire area detection: OD-CNN

Experiments showed that the 3rd method worked best!

Embedded Applications for Drones, Cloud, Smart phones, ...

CMSR Machine Learning Studio can generate deep learning model program source codes in Java, C/C++ and SWIFT. This code can be embedded into drone's firmware, cloud and smart phone applications, etc. Any system that is based on IEEE 754 floating point standards will be compatible.

For more information ...

Please contact us.