Table 6 The ablation experiments.
From: YOLOFM: an improved fire and smoke object detection algorithm based on YOLOv5n
Model | Num | Class | Precision | Recall | mAP50 | mAP50-95 | FPS | Params (MB) | GFLOPs(G) |
|---|---|---|---|---|---|---|---|---|---|
YOLOv5n | N1 | All | 91.8 | 90.9 | 95.3 | 66.8 | 80.90 | 6.72 | 4.1 |
Fire | 89.3 | 89.6 | 94.2 | 62.0 | |||||
Smoke | 94.2 | 92.1 | 96.4 | 71.5 | |||||
FocalNext | N2 | All | 92.5 | 92.5 | 95.8 | 67.8 | 77.94 | 6.64 | 4.2 |
Fire | 90.5 | 93.9 | 95.1 | 64.5 | |||||
Smoke | 94.5 | 91.1 | 96.5 | 71.1 | |||||
FocalNext+QAHARep-FPN | N3 | All | 93.3 | 93.3 | 96.8 | 72.3 | 67.98 | 8.55 | 5.7 |
Fire | 91.6 | 92.6 | 95.6 | 66.3 | |||||
Smoke | 94.9 | 94.0 | 98.0 | 78.3 | |||||
FocalNext+QAHARep -FPN +NADH | N4 | All | 94.2 | 94.2 | 97.2 | 73.6 | 66.58 | 13.87 | 6.0 |
Fire | 92.5 | 93.4 | 96.0 | 67.9 | |||||
Smoke | 95.9 | 94.9 | 98.3 | 79.4 | |||||
FocalNext+QAHARep -FPN+NADH+Focal-SIoU | N5 | All | 94.9 | 94.8 | 97.5 | 74.7 | 66.58 | 13.87 | 6.4 |
Fire | 93.5 | 94.2 | 96.3 | 68.8 | |||||
Smoke | 96.2 | 95.4 | 98.7 | 80.5 |
