Image Classification On Imagenet Real
평가 지표
Accuracy
Params
평가 결과
이 벤치마크에서 각 모델의 성능 결과
비교 표
| 모델 이름 | Accuracy | Params |
|---|---|---|
| large-scale-learning-of-general-visual | 90.54% | 928M |
| the-effectiveness-of-mae-pre-pretraining-for | 91.1% | - |
| resmlp-feedforward-networks-for-image | 85.6% | 45M |
| compounding-the-performance-improvements-of | 87.82% | - |
| resmlp-feedforward-networks-for-image | - | - |
| large-scale-learning-of-general-visual | 89.02% | - |
| model-soups-averaging-weights-of-multiple | 91.20% | 1843M |
| incorporating-convolution-designs-into-visual | 83.6% | - |
| tokenlearner-what-can-8-learned-tokens-do-for | 91.05% | 460M |
| meta-pseudo-labels | 91.02% | - |
| vitaev2-vision-transformer-advanced-by | 91.2% | 644M |
| model-soups-averaging-weights-of-multiple | 91.03% | 2440M |
| fixing-the-train-test-resolution-discrepancy | 89.73% | 829M |
| levit-a-vision-transformer-in-convnet-s | 87.5% | - |
| deit-iii-revenge-of-the-vit | - | - |
| volo-vision-outlooker-for-visual-recognition | 90.6% | - |
| resmlp-feedforward-networks-for-image | 84.6% | 15M |
| learning-transferable-architectures-for | 87.56% | - |
| compounding-the-performance-improvements-of | 88.65% | - |
| training-data-efficient-image-transformers | 82.1% | 5M |
| resnet-strikes-back-an-improved-training | 85.7% | 25M |
| very-deep-convolutional-networks-for-large | 80.60% | - |
| when-vision-transformers-outperform-resnets | 86.4% | - |
| levit-a-vision-transformer-in-convnet-s | 86.9% | - |
| sequencer-deep-lstm-for-image-classification | 87.9 | - |
| revisiting-weakly-supervised-pre-training-of | 90.7% | - |
| deit-iii-revenge-of-the-vit | - | - |
| volo-vision-outlooker-for-visual-recognition | 90.5% | - |
| fixing-the-train-test-resolution-discrepancy-2 | 90.9% | 480M |
| fixing-the-train-test-resolution-discrepancy-2 | 90.0% | 87M |
| incorporating-convolution-designs-into-visual | 88.1% | - |
| levit-a-vision-transformer-in-convnet-s | 85.6% | - |
| cvt-introducing-convolutions-to-vision | 90.6% | - |
| levit-a-vision-transformer-in-convnet-s | 82.6% | - |
| incorporating-convolution-designs-into-visual | 87.3% | - |
| scaling-vision-transformers | 90.81% | - |
| training-data-efficient-image-transformers | 88.7% | 86M |
| resmlp-feedforward-networks-for-image | 85.3% | 30M |
| deit-iii-revenge-of-the-vit | - | - |
| going-deeper-with-image-transformers | 90.2% | - |
| when-vision-transformers-outperform-resnets | 85.2% | - |
| revisiting-a-knn-based-image-classification | 84% | - |
| the-effectiveness-of-mae-pre-pretraining-for | 90.8% | - |
| mlp-mixer-an-all-mlp-architecture-for-vision | 87.86% | 409M |
| the-effectiveness-of-mae-pre-pretraining-for | 90.9% | - |
| training-data-efficient-image-transformers | 86.8% | 22M |
| self-training-with-noisy-student-improves | 90.55% | 480M |
| meta-pseudo-labels | 91.12% | - |
| when-vision-transformers-outperform-resnets | 84.4% | - |
| mlp-mixer-an-all-mlp-architecture-for-vision | 90.18% | 409M |
| levit-a-vision-transformer-in-convnet-s | 85.8% | - |
| learning-transferable-architectures-for | 81.15% | - |
| model-soups-averaging-weights-of-multiple | 91.78% | - |
| very-deep-convolutional-networks-for-large | 79.01% | - |
| imagenet-classification-with-deep | 62.88% | - |
| training-data-efficient-image-transformers | 89.3% | 86M |
| vision-models-are-more-robust-and-fair-when | 89.8% | 10000M |