Image-only low-shot recognition on ShapeNet LS
625-way 1-shot AccuracyDOPE
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| DOPESupervision=self-supervised, Training Dataset=ABC2022.11 | 62 | 73.55 | 47.93 | 60.72 | |
| VISPE++Supervision=self-supervised, Base SSL Framework=MoCoV2, Training Dataset=ABC2022.11 | 58.99 | 72.26 | 44.96 | 59.02 | |
| DOPESupervision=self-supervised2022.11 | 58.64 | 70.43 | 44.26 | 56.07 | |
| VISPE++Supervision=self-supervised, Base SSL Framework=SimSiam, Training Dataset=ABC2022.11 | 58.48 | 72.19 | 44.59 | 59.17 | |
| SimpleShotSupervision=supervised2022.11 | 58.07 | 70.06 | 43 | 55.96 | |
| RFSSupervision=supervised2022.11 | 57.93 | 73.23 | 43.08 | 59.71 | |
| FEATSupervision=supervised2022.11 | 57.83 | 72.41 | 42.92 | 58.95 | |
| VISPESupervision=self-supervised2022.11 | 57.69 | 68.65 | 42.43 | 54 | |
| VISPE++Supervision=self-supervised, Base SSL Framework=MoCoV22022.11 | 55.83 | 69.11 | 41.14 | 54.86 | |
| VISPE++Supervision=self-supervised, Base SSL Framework=SimSiam2022.11 | 53.22 | 66.75 | 39.12 | 52.25 |