ImageNet-C
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Image Classification | ImageNet-C Severity 5 (test) | Mean Error Rate (Severity 5)13.01 | 132 | |
| Image Classification | ImageNet-C (test) | Defocus Blur Acc100 | 125 | |
| Image Classification | ImageNet-C level 5 | Avg Top-1 Acc (ImageNet-C L5)86.2 | 110 | |
| Image Classification | ImageNet-C 1.0 (test) | Accuracy (Average)68.8 | 53 | |
| Online Continual Test-time Adaptation | ImageNet-C Severity 5 (test) | Accuracy (Gaussian Noise, ImageNet-C S5)63.4 | 47 | |
| Test-Time Adaptation | ImageNet-C level 5 (5k) | Accuracy (Gaussian Noise)61.9 | 47 | |
| Attack on Test-Time Adaptation | ImageNet-C ResNet50 (test) | Baseline Accuracy (BA)50.9 | 42 | |
| Image Classification | ImageNet-C 1.0 (val) | Corruption Error27.1 | 40 | |
| Image Classification | ImageNet-C severity level 5, continual setting | Gaussian Blur62 | 36 | |
| Image Classification | ImageNet-C Severity 5 Episodic (val) | Accuracy (Gaussian Noise)62 | 36 | |
| Image Classification | ImageNet-C | Gauss Error15.8 | 36 | |
| Image Classification | ImageNet-C blind-spot subset level 5 | Accuracy (N)53 | 35 | |
| Image Classification | ImageNet-C level 3 (test) | Acc (Brightness)81.9 | 34 | |
| Image Classification | ImageNet-C JPEG Corruptions | Top-1 Accuracy75.4 | 24 | |
| Image Classification | ImageNet-C Gaussian Noise | Top-1 Accuracy65.2 | 24 | |
| Image Classification | ImageNet-C | mCE28.3 | 20 | |
| Test-time adaptation | ImageNet-C γ = 10⁻¹ (test) | Accuracy (ImageNet-C, γ=10⁻¹)38.65 | 18 | |
| Transferability Estimation | ImageNet-C Severity 3 1.0 | Mean Correlation Improvement (MCI)22.6 | 16 | |
| Transferability Estimation | ImageNet-C Severity 1 1.0 | Mean Correlation Improvement (MCI)20.61 | 16 | |
| Image Classification | ImageNet-C level 5 (5k) | Accuracy (Gaussian Noise)70.1 | 16 | |
| Image Classification | ImageNet-C with Places365-C outliers 1.0 (test) | Accuracy46.9 | 14 | |
| Adaptation with outlier datasets | ImageNet-C with Textures-C outliers 1.0 (test) | Accuracy46.8 | 14 | |
| Adaptation with outlier datasets | ImageNet-C with Places365-C outliers 1.0 (test) | Accuracy0.469 | 14 | |
| Indiscriminate and Instant Attack | ImageNet-C (test) | Error Rate64.22 | 14 | |
| Image Classification | ImageNet-C severity level 5 1.0 (test) | Average Error Rate54.4 | 14 |
Showing 25 of 48 rows