Table 1 Quantitative Comparison on DeepShip Dataset.
Type | Method | Venue | Sup. | Self-sup. | Acc. | Prec. | Rec. | F1 |
|---|---|---|---|---|---|---|---|---|
Traditional ML | SVM | ESA (2021) | \(\checkmark\) | Â | 72.24 | 72.49 | 72.08 | 72.28 |
RF | \(\checkmark\) | Â | 69.71 | 69.79 | 69.86 | 69.82 | ||
KNN | \(\checkmark\) | Â | 62.71 | 63.61 | 63.10 | 63.35 | ||
Deep Learning | SCAE | ESA (2021) | \(\checkmark\) | Â | 77.53 | 77.75 | 77.41 | 77.58 |
Residual CNN | \(\checkmark\) | Â | 76.98 | 77.05 | 76.81 | 76.92 | ||
Inception | \(\checkmark\) | Â | 76.16 | 76.03 | 76.12 | 76.08 | ||
DNN | \(\checkmark\) | Â | 73.11 | 72.98 | 73.08 | 73.03 | ||
SSAST | AAAI 2022 | Â | \(\checkmark\) | 77.70 | 78.13 | 78.25 | 78.19 | |
AudioMAE | NeurIPS 2022 | Â | \(\checkmark\) | 76.66 | 85.54 | 79.00 | 82.14 | |
SSLM-M | JASA 2023 | Â | \(\checkmark\) | 80.22 | 80.81 | 79.94 | 80.07 | |
SNANet | AA 2023 | \(\checkmark\) | Â | 78.25 | 79.55 | 79.39 | 79.16 | |
MIXUP | JSTARS 2023 | Â | \(\checkmark\) | 86.33 | 85.72 | 82.91 | 84.29 | |
TR-Tral | TASLP 2024 | Â | \(\checkmark\) | 87.26 | 87.45 | 87.80 | 87.50 | |
Ours | — |  | \(\checkmark\) | 88.48 | 89.42 | 89.41 | 89.41 |
