Fine-grained Attribute Recognition on SLIP-RS Constructed Benchmark Ship
27.82Purpose mAPSLIP-RS
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| SLIP-RSBackbone=ConvNeXT-L, Training data=RS-Attribute-15M2026.05 | 27.82 | 48.03 | 27.61 | |
| SLIP-RSBackbone=ConvNeXT-T, Training data=RS-Attribute-15M2026.05 | 27.26 | 44.52 | 17.61 | |
| SLIP-RSBackbone=ConvNeXT-T, Training data=RS-Attribute-O2026.05 | 20.4 | 34 | 18.33 | |
| OpenRSDBackbone=RTMDet-L, Training data=ORSD+2026.05 | 15.41 | 12.36 | 15.67 | |
| LAE-DINOBackbone=Swin-T, Training data=LAE-1M2026.05 | 8.41 | 5.47 | 9.84 |