Wildfire And Smoke Classification

The Wildfire and Smoke Classification pretrained model enables the identification of wildfires and the early signs of smoke that may lead to a wildfire. The model uses imagery from aerial drones or ground-based camera systems and can classify wildfires and smoke across various settings and lighting conditions. Identifying wildfire threats early enables quicker intervention and more effective resource allocation, leading to improved wildfire management and response, and helping to mitigate the impact of wildfires on communities and ecosystems.

Using the model
Follow the guide to use the model. Before using this model, ensure that the supported deep learning libraries are installed. For more details, check Deep Learning Libraries Installer for ArcGIS.

Fine-tuning the model
This model cannot be fine-tuned using ArcGIS.

Input
8 bit, 3-band (RGB) image. This model is expected to work on high resolution drone imagery in the form of a raster, mosaic dataset, or image service.

Output
Classified image of the scene into one of the classes:  fire or nofire.

Applicable geographies
This model is expected to work well in all regions globally. However, results can vary for images that are statistically dissimilar to training data.

Model architecture
This model uses the ResNet50 model architecture implemented in ArcGIS API for Python.

Accuracy metrics
This model has an overall accuracy of 96.28 percent. 

Limitations
This model may misclassify fires in urban environments. Therefore, it is recommended for use primarily in wild forest areas.

Sample results
Here are a few results from the model.