Table 2 Zero-shot recognition results (± standard deviation) of the SAEVRT method compared with those of the SAE-based methods (\(\%\)).
From: Enhancing zero-shot scene recognition through semantic autoencoders and visual relation transfer
Method | Scene15 | MIT67 | UCM21 | NWPU45 |
|---|---|---|---|---|
SAE | 56.87 ± 8.77 | 63.28 ± 9.65 | 49.50 ± 8.42 | 44.81 ± 6.73 |
DDSA | 56.89 ± 8.69 | 64.18 ± 9.92 | 51.71 ± 7.65 | 47.50 ± 6.05 |
CAE | 59.20 ± 8.95 | 65.18 ± 10.19 | 54.89 ± 9.28 | 49.13 ± 6.68 |
DIPL | 61.43 ± 9.14 | 65.97 ± 10.23 | 56.54 ± 9.95 | 50.42 ± 5.49 |
SAEVRT | 63.77 ± 9.71 | 67.75 ± 10.35 | 58.68 ± 10.04 | 53.26 ± 6.11 |
