Table 11 Simulation results - learning rate impact on NASA dataset.
From: Real time fire and smoke detection using vision transformers and spatiotemporal learning
Learning rate | Proposed hybrid model | ResNet50 | VGG16 | LSTM | 3D-CNNs | Hybrid ResNet50 + LSTM | Hybrid VGG16 + 3D-CNN |
|---|---|---|---|---|---|---|---|
0.001 | 99.2% | 90.5% | 87.6% | 91.3% | 94.7% | 95.8% | 95.2% |
0.01 | 98.7% | 89.4% | 86.5% | 90.4% | 93.6% | 94.7% | 94.1% |
0.1 | 96.5% | 85.2% | 82.8% | 87.5% | 91.2% | 93.1% | 92.5% |
0.0001 | 99.0% | 89.8% | 86.2% | 89.8% | 94.1% | 94.5% | 93.9% |
