ImageNet-C
Benchmarks
TaskDatasetSOTA resultTrendResults
ImageNet-C Severity 5 (test)
13.01Mean Error Rate (Severity 5)
216
ImageNet-C (test)
100Defocus Blur Acc
139
ImageNet-C level 5
86.2Avg Top-1 Acc (ImageNet-C L5)
110
ImageNet-C 1.0 (test)
68.8Accuracy (Average)
56
ImageNet-C Severity 5 (test)
63.4Accuracy (Gaussian Noise, ImageNet-C S5)
47
ImageNet-C level 5 (5k)
61.9Accuracy (Gaussian Noise)
47
ImageNet-C ResNet50 (test)
50.9Baseline Accuracy (BA)
42
ImageNet-C 1.0 (val)
27.1Corruption Error
40
ImageNet-C severity level 5, continual setting
62Gaussian Blur
36
ImageNet-C Severity 5 Episodic (val)
62Accuracy (Gaussian Noise)
36
ImageNet-C
15.8Gauss Error
36
ImageNet-C Severity 3 1.0
74.8Accuracy (Gauss Corruption)
35
ImageNet-C blind-spot subset level 5
53Accuracy (N)
35
ImageNet-C level 3 (test)
81.9Acc (Brightness)
34
ImageNet-C
77.8Top-1 Accuracy
31
ImageNet-C JPEG Corruptions
75.4Top-1 Accuracy
24
ImageNet-C Gaussian Noise
65.2Top-1 Accuracy
24
ImageNet-C γ = 10⁻¹ (test)
38.65Accuracy (ImageNet-C, γ=10⁻¹)
18
ImageNet-C Severity 3 1.0
22.6Mean Correlation Improvement (MCI)
16
ImageNet-C Severity 1 1.0
20.61Mean Correlation Improvement (MCI)
16
ImageNet-C level 5 (5k)
70.1Accuracy (Gaussian Noise)
16
ImageNet-C with Places365-C outliers 1.0 (test)
46.9Accuracy
14
ImageNet-C with Textures-C outliers 1.0 (test)
46.8Accuracy
14
ImageNet-C with Places365-C outliers 1.0 (test)
0.469Accuracy
14
ImageNet-C (test)
64.22Error Rate
14