ImageNet
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
ImageNet 256x256 (train)
2.15FID
21
ImageNet 1k (val)
71Top-1 Accuracy
21
ImageNet New classes
73.92New Accuracy (%)
21
ImageNet 1k (val)
72.35Accuracy
21
ImageNet 64x64 (train)
1.23FID
21
ImageNet-100 (test)
86.5Accuracy (All)
21
ImageNet ID
82.9Accuracy
21
ImageNet-S300 (val)
35.1mIoU
21
ImageNet-S (val)
63mIoU
21
ImageNet 1k (val)
76.89Top-1 Accuracy
21
ImageNet-LT 1.0 (test)
57.1Top-1 Accuracy (Overall)
21
ImageNet-1000 10 steps
90.6Top-5 Acc (Last)
21
mini-ImageNet-Red 80% noise
51.2Accuracy
21
mini-ImageNet-Red 40% noise
65.6Accuracy
21
mini-ImageNet-Red 20% noise
69Accuracy
21
S-Tiny-ImageNet
82.04Task-IL Accuracy
21
ImageNet 1k (v2)
84Top-1 Accuracy
21
Tiny-ImageNet (val)
62.35Average Accuracy
21
ImageNet127
75.8Accuracy
21
ImageNet Real re-assessed (val)
88.1Real Top-1 Acc
21
ImageNet-1K 10
55.6Overall Accuracy
21
tiered-ImageNet (test)
71.71Accuracy
21
ImageNet-GLT GBL 1k (test)
60.79Accuracy (Manyc)
21
ImageNet 1% labeled 1.0
84.6Accuracy
21
ImageNet
82.4Accuracy
20