Table 2 Comparisons of recall (rec) and F1-score (F1S) on DarkNet-53 and YOLOv3
From: Bridge-over-water detection via modulated deformable convolution and attention mechanisms
Backbones | \(\mathrm Rec\)(%) | \(\mathrm Rec^{s}\)(%) | \(\mathrm Rec^{m}\)(%) | \(\mathrm Rec^{l}\)(%) | \(\mathrm F1S\)(%) |
|---|---|---|---|---|---|
Plain | 60.4 | 43.6 | 68.8 | 67.6 | 51.7 |
Plain+Mdconv | 64.5 | 45.2 | 70.2 | 80.2 | 58.7 |
Plain+Mdconv+CBAM | 68.2 | 51.3 | 72.4 | 83.1 | 63.8 |
Pain+Mdconv+WCSA(CBAM) | 68.0 | 52.3 | 71.3 | 83.2 | 62.5 |
Plain+Mdconv+SCSA(CBAM) | 68.6 | 51.3 | 72.4 | 84.7 | 62.7 |
Plain+Mdconv+BAM | 68.1 | 51.7 | 72.2 | 82.6 | 62.4 |
Plain+Mdconv+WCSA(BAM) | 68.1 | 50.6 | 72.2 | 84.2 | 63.0 |
Plain+Mdconv+SCSA(BAM) | 68.5 | 52.3 | 72.3 | 83.5 | 63.1 |