Fig. 20 | Scientific Reports

Fig. 20

From: Hybrid learning framework for synergistic fusion of SAR and optical UAV data in wildfire surveillance

Fig. 20

This plot compares the dynamics of training loss of various UAV-based wildfire detection models. The DualSegFormer (UAV) model presents the steadiest and fastest fall in both Dice Loss and Custom Loss, implying better segmentation quality and quicker convergence. Models such as YOLOv10 and YOLOv9 present slower advancement and greater terminal loss values, implying relatively constrained segmentation accuracy in wildfire environments.

Back to article page