Semi-Supervised Image Classification 벤치마크
Semi-Supervised Image Classification on SVHN (1000 Labels, ImageNet-100 Unlabeled)
Accuracy
- 2022-03-04 — CCSSL: Accuracy 88.6
- 2023-06-02 — UnMixMatch: Accuracy 91.03
| Rank | Model | Accuracy | Paper | Code | Year |
|---|---|---|---|---|---|
| 1 | UnMixMatch | 91.03 | Scaling Up Semi-supervised Learning with Unconstrained Unlabelled Data | shuvenduroy/unmixmatch | 2023 |
| 2 | CCSSL | 88.6 | Class-Aware Contrastive Semi-Supervised Learning | tencentyouturesearch/classification-semicls | 2022 |
| 3 | UnMixMatch | 91.03 | Scaling Up Semi-supervised Learning with Unconstrained Unlabelled Data | shuvenduroy/unmixmatch | 2023 |
| 4 | CCSSL | 88.6 | Class-Aware Contrastive Semi-Supervised Learning | tencentyouturesearch/classification-semicls | 2022 |