Research Article
[Retracted] Relieving the Incompatibility of Network Representation and Classification for Long-Tailed Data Distribution
Table 6
Long-tailed classification results on ImageNet LT.
| Model | Many-shot | Medium-shot | Few-shot | Acc | ResNet-10 | >100 | ≤100 and >20 | ≤20 |
| Plain model | 56.8 | 25.7 | 3.6 | 34.6 | Many-shot model | 57.9 | — | — | — | Lifted loss [41] | 35.8 | 30.4 | 17.9 | 30.8 | Focal loss [13] | 36.4 | 29.9 | 16 | 30.5 | Range loss [14] | 35.8 | 30.3 | 17.6 | 30.7 | FsLwf [42] | 40.9 | 22.1 | 15 | 28.4 | OLTR [10] | 43.2 | 35.1 | 18.5 | 35.6 | BALMS [43] | 50.3 | 39.5 | 25.3 | 41.8 | Decouple (cRT) [12] | 52.3 | 39.5 | 23.2 | 42.1 | Decouple (-normalized) [12] | 51.9 | 38.3 | 22.5 | 40.6 | Ours with OLTR | 54.0 | 34.1 | 17.4 | 39.2 | Ours with decouple | 54.9 | 39.6 | 23.4 | 43.9 |
|
|
Results directly copied from Ref. [ 10]. Results reproduced with author’s code. |