List your company
ResearchFormerly WizWand

Test-time adaptation on ImageNet-C (test)

64.5AccuracyZeroSiam
11.35625.15338.9552.747Sep 27, 2025
Updated 7mo ago

Evaluation results

MethodLinks
ZeroSiamBackbone=ViT-Base, Mem. (MB)=7,806, Time (s)=113, Adaptation scenario=blind-spot adaptation, Batch size=64, Hardware=A100 GPU2025.0964.5
SPABackbone=ViT-Base, Mem. (MB)=28,147, Time (s)=275, Adaptation scenario=blind-spot adaptation, Batch size=64, Hardware=A100 GPU2025.0960.3
REMBackbone=ViT-Base, Mem. (MB)=27,438, Time (s)=346, Adaptation scenario=blind-spot adaptation, Batch size=64, Hardware=A100 GPU2025.0956.9
TTEBackbone=ViT-Base, Mem. (MB)=8,574, Time (s)=164, Adaptation scenario=blind-spot adaptation, Batch size=64, Hardware=A100 GPU2025.0955.2
ZeroSiamBackbone=ResNet50-GN, Mem. (MB)=5,584, Time (s)=89, Adaptation scenario=blind-spot adaptation, Batch size=64, Hardware=A100 GPU2025.0946.2
SPABackbone=ResNet50-GN, Mem. (MB)=21,327, Time (s)=220, Adaptation scenario=blind-spot adaptation, Batch size=64, Hardware=A100 GPU2025.0945
TTEBackbone=ResNet50-GN, Mem. (MB)=6,303, Time (s)=136, Adaptation scenario=blind-spot adaptation, Batch size=64, Hardware=A100 GPU2025.0940.8
TentBackbone=ViT-Base, Mem. (MB)=7,760, Time (s)=113, Adaptation scenario=blind-spot adaptation, Batch size=64, Hardware=A100 GPU2025.0930.7
TentBackbone=ResNet50-GN, Mem. (MB)=5,519, Time (s)=89, Adaptation scenario=blind-spot adaptation, Batch size=64, Hardware=A100 GPU2025.0913.4