Self Supervised Image Classification On 1
评估指标
Number of Params
Top 1 Accuracy
评测结果
各个模型在此基准测试上的表现结果
比较表格
| 模型名称 | Number of Params | Top 1 Accuracy |
|---|---|---|
| dinov2-learning-robust-visual-features | 1100M | 88.9% |
| efficient-self-supervised-vision-transformers | 87M | 83.9% |
| ibot-image-bert-pre-training-with-online | 307M | 84.8% |
| architecture-agnostic-masked-image-modeling | - | 80.4% |
| momentum-contrast-for-unsupervised-visual | - | 77.0% |
| ibot-image-bert-pre-training-with-online | 307M | 86.6% |
| masked-image-residual-learning-for-scaling-1 | 341M | 86.2% |
| simmim-a-simple-framework-for-masked-image | 658M | 87.1% |
| divide-and-contrast-self-supervised-learning | - | 78.2% |
| architecture-agnostic-masked-image-modeling | - | 84.5% |
| improving-visual-representation-learning | 307M | 88.6% |
| architecture-agnostic-masked-image-modeling | - | 82.4% |
| dinov2-learning-robust-visual-features | 1100M | 88.5% |
| designing-bert-for-convolutional-networks | 60M | 82.7% |
| designing-bert-for-convolutional-networks | 89M | 84.8% |
| ibot-image-bert-pre-training-with-online | 85M | 84.0% |
| beit-bert-pre-training-of-image-transformers | 307M | 86.3% |
| masked-feature-prediction-for-self-supervised | 307M | 85.7% |
| architecture-agnostic-masked-image-modeling | - | 84.2% |
| unsupervised-learning-of-visual-features-by | 193M | 82.0% |
| masked-autoencoders-are-scalable-vision | - | 86.9% |
| designing-bert-for-convolutional-networks | 198M | 86.0% |
| designing-bert-for-convolutional-networks | 26M | 80.6% |
| emerging-properties-in-self-supervised-vision | 85M | 82.8% |
| towards-sustainable-self-supervised-learning | - | 86.5% |
| big-self-supervised-models-are-strong-semi | 795M | 83.1% |
| masked-autoencoders-are-scalable-vision | 632M | 87.8% |
| designing-bert-for-convolutional-networks | 65M | 83.1% |
| architecture-agnostic-masked-image-modeling | - | 78.9% |
| augmenting-sub-model-to-improve-main-model | 87M | 83.9% |
| simmim-a-simple-framework-for-masked-image | 88M | 84.0% |
| momentum-contrast-for-unsupervised-visual | - | 77.3% |
| unifying-architectures-tasks-and-modalities | 473M | 85.6% |
| architecture-agnostic-masked-image-modeling | - | 78.8% |
| simmim-a-simple-framework-for-masked-image | 85M | 83.8% |
| an-empirical-study-of-training-self | 86M | 83.2% |
| exploring-target-representations-for-masked | 632M | 88.0% |
| context-autoencoder-for-self-supervised | 307M | 86.3% |
| mc-beit-multi-choice-discretization-for-image | 86M | 84.1% |
| designing-bert-for-convolutional-networks | 50M | 84.1% |
| a-simple-framework-for-contrastive-learning | - | 77.2% |
| bootstrapped-masked-autoencoders-for-vision | 307M | 85.9% |
| self-supervised-pretraining-of-visual | 1.3B | 84.2% |
| leveraging-large-scale-uncurated-data-for | 138M | 74.9% |
| architecture-agnostic-masked-image-modeling | - | 80.5% |
| unsupervised-learning-of-visual-features-by | 182M | 77.8% |
| simmim-a-simple-framework-for-masked-image | 197M | 85.4% |
| augmenting-sub-model-to-improve-main-model | 304M | 86.1% |
| designing-bert-for-convolutional-networks | 44M | 82.2% |
| peco-perceptual-codebook-for-bert-pre | 632M | 88.3% |
| mugs-a-multi-granular-self-supervised | 21M | 82.6% |
| student-collaboration-improves-self | - | 83.2% |
| an-empirical-study-of-training-self | 304M | 84.1% |
| self-supervised-pretraining-of-visual | 693M | 83.8% |
| ibot-image-bert-pre-training-with-online | 307M | 87.8% |
| designing-bert-for-convolutional-networks | 198M | 85.4% |
| improving-visual-representation-learning | 307M | 88.1% |
| masked-image-residual-learning-for-scaling-1 | 96M | 84.8% |
| architecture-agnostic-masked-image-modeling | - | 82.2% |
| ibot-image-bert-pre-training-with-online | 85M | 84.4% |
| augmenting-sub-model-to-improve-main-model | 632M | 87.2% |
| vision-models-are-more-robust-and-fair-when | 10000M | 85.8% |
| mugs-a-multi-granular-self-supervised | 85M | 84.3% |
| mugs-a-multi-granular-self-supervised | 307M | 85.2% |
| beit-bert-pre-training-of-image-transformers | 86M | 84.6% |