It seems that this project is limited to widely be applied to other work(\eg, wildfire detection, indoor wildfire detection)

[xbyang1000]

This work aims to annotate fire pixel easier, e.g., fire pixels should be paid more attention than non-fire ones. Due to following KNN classification, the number of classes and the number of training pixels in each class cannot be too large.