Performance Evaluation of Deep Learning Algorithm for Forest Fire Detection
Parinya Suwansrikham, Phudinan Singkhamfu · 2023
The forest fire caused further destruction to the woodland region. It must be monitored and detected as early as possible. In the study, the pre-trained deep learning performance of Inception, ResNet, and ViT was compared for the categorization of photos of forest fires. These models' performances are evaluated with one another. The findings from the study indicate that the ViT model has the best accuracy at 88.03%. The ResNet-26, ResNet-50, Inception v3, and Inception v4 models have an accuracy of 86.66%, 85.64%, 83.25%, and 80.86%, respectively.