Image Classification on ImageNet (Top1 Accuracy and Search Cost)
82.4Top-1 AccuracyBaseline
Evaluation results
| Method | Links | ||
|---|---|---|---|
| BaselineAuthorized Target Model=None, Evaluated Model=MambaVision-T, PSNR=20 dB2025.10 | 82.4 | — | |
| BaselineAuthorized Target Model=None, Evaluated Model=ViT-B, PSNR=20 dB2025.10 | 81.6 | — | |
| NTE (ViT-B)Authorized Target Model=ViT-B, Evaluated Model=ViT-B, PSNR=20 dB2025.10 | 81.3 | — | |
| NTE (MambaVision-T)Authorized Target Model=MambaVision-T, Evaluated Model=MambaVision-T, PSNR=20 dB2025.10 | 81 | — | |
| BaselineAuthorized Target Model=None, Evaluated Model=SwinV2-T, PSNR=20 dB2025.10 | 80.9 | — | |
| BaselineAuthorized Target Model=None, Evaluated Model=ResNet-50, PSNR=20 dB2025.10 | 80.3 | — | |
| NTE (ResNet-50)Authorized Target Model=ResNet-50, Evaluated Model=ResNet-50, PSNR=20 dB2025.10 | 80.2 | — | |
| BaselineAuthorized Target Model=None, Evaluated Model=DeiT-B, PSNR=20 dB2025.10 | 79.9 | — | |
| NTE (DeiT-B)Authorized Target Model=DeiT-B, Evaluated Model=DeiT-B, PSNR=20 dB2025.10 | 79.3 | — | |
| NAR-FormerMACs=571M2022.11 | 76.9 | 2 | |
| NAR-FormerMACs=378M2022.11 | 76.36 | 1.63 | |
| OFAMACs=230M2022.11 | 76 | 40 | |
| NTE (SwinV2-T)Authorized Target Model=SwinV2-T, Evaluated Model=SwinV2-T, PSNR=20 dB2025.10 | 71.7 | — | |
| NTE (SwinV2-T)Authorized Target Model=SwinV2-T, Evaluated Model=MambaVision-T, PSNR=20 dB2025.10 | 12.9 | — | |
| NTE (SwinV2-T)Authorized Target Model=SwinV2-T, Evaluated Model=ResNet-50, PSNR=20 dB2025.10 | 9.1 | — | |
| NTE (SwinV2-T)Authorized Target Model=SwinV2-T, Evaluated Model=DeiT-B, PSNR=20 dB2025.10 | 7.4 | — | |
| NTE (MambaVision-T)Authorized Target Model=MambaVision-T, Evaluated Model=ResNet-50, PSNR=20 dB2025.10 | 5.8 | — | |
| NTE (SwinV2-T)Authorized Target Model=SwinV2-T, Evaluated Model=ViT-B, PSNR=20 dB2025.10 | 4.3 | — | |
| NTE (MambaVision-T)Authorized Target Model=MambaVision-T, Evaluated Model=SwinV2-T, PSNR=20 dB2025.10 | 1.5 | — | |
| NTE (DeiT-B)Authorized Target Model=DeiT-B, Evaluated Model=MambaVision-T, PSNR=20 dB2025.10 | 1 | — | |
| NTE (MambaVision-T)Authorized Target Model=MambaVision-T, Evaluated Model=DeiT-B, PSNR=20 dB2025.10 | 0.7 | — | |
| NTE (ResNet-50)Authorized Target Model=ResNet-50, Evaluated Model=SwinV2-T, PSNR=20 dB2025.10 | 0.1 | — | |
| NTE (ResNet-50)Authorized Target Model=ResNet-50, Evaluated Model=DeiT-B, PSNR=20 dB2025.10 | 0.1 | — | |
| NTE (ResNet-50)Authorized Target Model=ResNet-50, Evaluated Model=MambaVision-T, PSNR=20 dB2025.10 | 0.1 | — | |
| NTE (ResNet-50)Authorized Target Model=ResNet-50, Evaluated Model=ViT-B, PSNR=20 dB2025.10 | 0 | — | |
| NTE (ViT-B)Authorized Target Model=ViT-B, Evaluated Model=ResNet-50, PSNR=20 dB2025.10 | 0 | — | |
| NTE (ViT-B)Authorized Target Model=ViT-B, Evaluated Model=SwinV2-T, PSNR=20 dB2025.10 | 0 | — | |
| NTE (ViT-B)Authorized Target Model=ViT-B, Evaluated Model=DeiT-B, PSNR=20 dB2025.10 | 0 | — | |
| NTE (ViT-B)Authorized Target Model=ViT-B, Evaluated Model=MambaVision-T, PSNR=20 dB2025.10 | 0 | — | |
| NTE (DeiT-B)Authorized Target Model=DeiT-B, Evaluated Model=ResNet-50, PSNR=20 dB2025.10 | 0 | — | |
| NTE (DeiT-B)Authorized Target Model=DeiT-B, Evaluated Model=ViT-B, PSNR=20 dB2025.10 | 0 | — | |
| NTE (DeiT-B)Authorized Target Model=DeiT-B, Evaluated Model=SwinV2-T, PSNR=20 dB2025.10 | 0 | — | |
| NTE (MambaVision-T)Authorized Target Model=MambaVision-T, Evaluated Model=ViT-B, PSNR=20 dB2025.10 | 0 | — |