# Long-tailed Recognition on ImageNet-LT Forward-LT (test)

Source: [Long-tailed Recognition on ImageNet-LT Forward-LT (test)](<https://www.sota2.com/research/sota/long-tailed-recognition-on-imagenet-lt-forward-lt-test>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-02-26T06:06:53.539Z

Evaluation of Long-tailed Recognition performance on the test split of the ImageNet-LT Forward-LT dataset, reporting Top-1 Accuracy segmented by Inverse Ranking (IR) groups (IR=50 down to IR=2).

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

| Scope | Recorded value |
| --- | --- |
| Task | Long-tailed Recognition |
| Dataset (including any stated split) | ImageNet-LT Forward-LT (test) |
| Metric | Top-1 Acc (IR=50) |
| Unit | PERCENT |
| Direction | Higher is better |
| Results loaded | 7 of 7 |
| Coverage | All results currently returned for this benchmark |

Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol.

## Metrics

- [Top-1 Acc (IR=50)](<https://www.sota2.com/research/sota/long-tailed-recognition-on-imagenet-lt-forward-lt-test.md?metric=Top-1%20Acc%20(IR%3D50)>): Higher is better; unit: PERCENT.

- [Top-1 Acc (IR=25)](<https://www.sota2.com/research/sota/long-tailed-recognition-on-imagenet-lt-forward-lt-test.md?metric=Top-1%20Acc%20(IR%3D25)>): Higher is better; unit: PERCENT.

- [Top-1 Acc (IR=10)](<https://www.sota2.com/research/sota/long-tailed-recognition-on-imagenet-lt-forward-lt-test.md?metric=Top-1%20Acc%20(IR%3D10)>): Higher is better; unit: PERCENT.

- [Top-1 Acc (IR=5)](<https://www.sota2.com/research/sota/long-tailed-recognition-on-imagenet-lt-forward-lt-test.md?metric=Top-1%20Acc%20(IR%3D5)>): Higher is better; unit: PERCENT.

- [Top-1 Acc (IR=2)](<https://www.sota2.com/research/sota/long-tailed-recognition-on-imagenet-lt-forward-lt-test.md?metric=Top-1%20Acc%20(IR%3D2)>): Higher is better; unit: PERCENT.

## Recorded results

| Method | Top-1 Acc (IR=50) | Result date | Report paper |
| --- | --- | --- | --- |
| SADE | 69.4 | Not recorded | 68fa6dc51e209131cea3c01a |
| RIDE | 67.6 | Not recorded | 68fa6dc51e209131cea3c01a |
| Softmax | 66.1 | Not recorded | 68fa6dc51e209131cea3c01a |
| LADE | 65.8 | Not recorded | 68fa6dc51e209131cea3c01a |
| LADE | 63.4 | Not recorded | 68fa6dc51e209131cea3c01a |
| BS | 63.2 | Not recorded | 68fa6dc51e209131cea3c01a |
| MISLAS | 61.6 | Not recorded | 68fa6dc51e209131cea3c01a |

### SADE

Top-1 Acc (IR=50): 69.4. Result ID: 698fdc723a6c314849fc5fc9.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6dc51e209131cea3c01a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6dc51e209131cea3c01a.md>)

[Implementation](<https://github.com/Vanint/SADE-AgnosticLT>)

Recorded evaluation configuration: {"prior":false}. This record does not assert independent reproduction.

### RIDE

Top-1 Acc (IR=50): 67.6. Result ID: 698fdc723a6c314849fc5fc8.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6dc51e209131cea3c01a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa6c7b1e209131cea3436b.md>)

[Implementation](<https://github.com/frank-xwang/RIDE-LongTailRecognition>)

Recorded evaluation configuration: {"prior":false}. This record does not assert independent reproduction.

### Softmax

Top-1 Acc (IR=50): 66.1. Result ID: 698fdc723a6c314849fc5fc3.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6dc51e209131cea3c01a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69b21e209131cea237df.md>)

[Implementation](<https://github.com/Jia-Research-Lab/MiSLAS>)

Recorded evaluation configuration: {"prior":false}. This record does not assert independent reproduction.

### LADE

Top-1 Acc (IR=50): 65.8. Result ID: 698fdc723a6c314849fc5fc7.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6dc51e209131cea3c01a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa698d1e209131cea228f0.md>)

[Implementation](<https://github.com/hyperconnect/LADE>)

Recorded evaluation configuration: {"prior":true}. This record does not assert independent reproduction.

### LADE

Top-1 Acc (IR=50): 63.4. Result ID: 698fdc723a6c314849fc5fc6.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6dc51e209131cea3c01a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa698d1e209131cea228f0.md>)

[Implementation](<https://github.com/hyperconnect/LADE>)

Recorded evaluation configuration: {"prior":false}. This record does not assert independent reproduction.

### BS

Top-1 Acc (IR=50): 63.2. Result ID: 698fdc723a6c314849fc5fc4.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6dc51e209131cea3c01a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa68c11e209131cea1e125.md>)

[Implementation](<https://github.com/jiawei-ren/BalancedMetaSoftmax>)

Recorded evaluation configuration: {"prior":false}. This record does not assert independent reproduction.

### MISLAS

Top-1 Acc (IR=50): 61.6. Result ID: 698fdc723a6c314849fc5fc5.

[Reporting paper](<https://www.sota2.com/research/paper/68fa6dc51e209131cea3c01a.md>)

[Method origin paper](<https://www.sota2.com/research/paper/68fa69b21e209131cea237df.md>)

[Implementation](<https://github.com/Jia-Research-Lab/MiSLAS>)

Recorded evaluation configuration: {"prior":false}. This record does not assert independent reproduction.

Page 1 of 1. This is an index page, not the full collection.