This article [J. of Electronic Imaging, 31(5), 053002 (2022) doi 10.1117/1.JEI.31.5.053002] was originally published on 1 September 2022 with an error in Table 2. The best results were mistakenly displayed in italics and the second-best results were displayed in bold. The table has been corrected to display the best results in bold and the second-best results in italics. The corrected table is below.
Table 2
The ablation study results of the different backbones on four benchmark datasets. ↑ indicates the higher the better, and vice versa. The best score of each indicator is highlighted in italics, and the second-best score is in bold.
Backbone | CHAMELEON | CAMO-test | COD10K-test | NC4K |
---|
Sα↑ | Em↑ | Fβω↑ | M↓ | Sα↑ | Em↑ | Fβω↑ | M↓ | Sα↑ | Em↑ | Fβω↑ | M↓ | Sα↑ | Em↑ | Fβω↑ | M↓ |
---|
ResNet1846 | 0.794 | 0.855 | 0.651 | 0.066 | 0.624 | 0.676 | 0.429 | 0.142 | 0.642 | 0.721 | 0.379 | 0.083 | 0.681 | 0.741 | 0.497 | 0.107 |
ResNet3446 | 0.811 | 0.876 | 0.675 | 0.059 | 0.625 | 0.687 | 0.431 | 0.142 | 0.644 | 0.722 | 0.380 | 0.085 | 0.684 | 0.747 | 0.500 | 0.107 |
ResNet5046 | 0.793 | 0.850 | 0.631 | 0.069 | 0.626 | 0.675 | 0.424 | 0.143 | 0.636 | 0.708 | 0.363 | 0.090 | 0.676 | 0.735 | 0.483 | 0.112 |
ResNet10146 | 0.774 | 0.823 | 0.605 | 0.077 | 0.614 | 0.662 | 0.407 | 0.148 | 0.638 | 0.704 | 0.363 | 0.091 | 0.676 | 0.729 | 0.478 | 0.113 |
ResNet15246 | 0.762 | 0.816 | 0.582 | 0.084 | 0.619 | 0.669 | 0.409 | 0.149 | 0.641 | 0.704 | 0.368 | 0.093 | 0.687 | 0.737 | 0.492 | 0.112 |
ResNeXt5047 | 0.785 | 0.850 | 0.625 | 0.067 | 0.626 | 0.690 | 0.427 | 0.145 | 0.646 | 0.714 | 0.379 | 0.088 | 0.688 | 0.746 | 0.501 | 0.108 |
ResNeXt10147 | 0.775 | 0.830 | 0.609 | 0.074 | 0.614 | 0.672 | 0.410 | 0.151 | 0.640 | 0.708 | 0.369 | 0.091 | 0.681 | 0.741 | 0.490 | 0.111 |
Wide_ResNet5048 | 0.800 | 0.854 | 0.655 | 0.065 | 0.622 | 0.678 | 0.418 | 0.145 | 0.643 | 0.714 | 0.378 | 0.087 | 0.683 | 0.742 | 0.496 | 0.109 |
Wide_ResNet10148 | 0.777 | 0.828 | 0.607 | 0.074 | 0.625 | 0.679 | 0.423 | 0.148 | 0.640 | 0.708 | 0.367 | 0.090 | 0.681 | 0.736 | 0.487 | 0.111 |
Res2Net5049 | 0.891 | 0.939 | 0.819 | 0.030 | 0.817 | 0.872 | 0.740 | 0.070 | 0.815 | 0.887 | 0.682 | 0.036 | 0.845 | 0.900 | 0.768 | 0.047 |
Res2Net10149 | 0.889 | 0.943 | 0.822 | 0.029 | 0.814 | 0.868 | 0.734 | 0.073 | 0.818 | 0.890 | 0.686 | 0.035 | 0.850 | 0.903 | 0.775 | 0.047 |
pvt_v2_b013 | 0.870 | 0.930 | 0.782 | 0.036 | 0.815 | 0.866 | 0.726 | 0.072 | 0.804 | 0.873 | 0.655 | 0.039 | 0.843 | 0.896 | 0.757 | 0.050 |
pvt_v2_b113 | 0.889 | 0.941 | 0.816 | 0.028 | 0.830 | 0.886 | 0.761 | 0.065 | 0.828 | 0.899 | 0.704 | 0.033 | 0.861 | 0.913 | 0.791 | 0.042 |
pvt_v2_b213 | 0.895 | 0.944 | 0.837 | 0.027 | 0.856 | 0.910 | 0.801 | 0.054 | 0.847 | 0.915 | 0.742 | 0.028 | 0.875 | 0.926 | 0.817 | 0.037 |
pvt_v2_b313 | 0.899 | 0.949 | 0.843 | 0.028 | 0.864 | 0.919 | 0.812 | 0.050 | 0.847 | 0.918 | 0.745 | 0.027 | 0.877 | 0.929 | 0.822 | 0.036 |
pvt_v2_b413 | 0.910 | 0.959 | 0.857 | 0.023 | 0.870 | 0.925 | 0.822 | 0.047 | 0.849 | 0.918 | 0.746 | 0.027 | 0.880 | 0.931 | 0.826 | 0.035 |
pvt_v2_b513 | 0.897 | 0.942 | 0.838 | 0.027 | 0.863 | 0.922 | 0.814 | 0.048 | 0.840 | 0.908 | 0.731 | 0.030 | 0.873 | 0.927 | 0.818 | 0.037 |
The article was corrected on 5 September 2022.