Table 6 Standard detection in clear daylight (NASA space apps challenge dataset).
From: Real time fire and smoke detection using vision transformers and spatiotemporal learning
Model | Accuracy (%) | Precision (%) | Recall (%) | F1-Score (%) | AUC-ROC (%) | Specificity (%) |
|---|---|---|---|---|---|---|
Proposed hybrid model | 99.2 | 99.3 | 99.0 | 99.1 | 99.5 | 99.4 |
ResNet50 | 90.5 | 89.7 | 91.1 | 90.4 | 92.3 | 88.6 |
VGG16 | 87.6 | 85.3 | 89.4 | 87.3 | 90.5 | 85.7 |
LSTM | 91.3 | 92.1 | 90.5 | 91.3 | 93.2 | 90.1 |
3D-CNNs | 94.7 | 95.0 | 94.4 | 94.7 | 96.1 | 93.2 |
Hybrid ResNet50 + LSTM | 95.8 | 95.5 | 96.2 | 95.8 | 96.8 | 95.3 |
Hybrid VGG16 + 3D-CNN | 95.2 | 94.9 | 95.5 | 95.2 | 96.3 | 94.7 |
