Research Article
Compressed Wavelet Tensor Attention Capsule Network
Table 2
Classification accuracies (%) of six texture classification methods on three texture datasets.
| Method | Datasets | CUReT | DTD | KTH-TIPS2-b |
| Wavelet CNNs | 78.95 ± 0.85 | 59.8 ± 0.92 | 74.2 ± 1.21 | T-CNN | 99.5 ± 0.4 | 55.8 ± 0.8 | 73.2 ± 2.2 | CapsNets | 92.4 ± 0.98 | 70.98 ± 1.03 | 73.83 ± 1.12 | FV-CNN | 95.7 ± 1.08 | 75.5 ± 0.8 | 81.5 ± 2.0 | SI-LCvMSP | 96.44 ± 0.75 | 76.7 ± 0.78 | 96.1 ± 1.02 | CWTACapsNet | 99.7 ± 0.22 | 81.52 ± 0.47 | 97.15 ± 0.55 |
|
|