Image Classification on ELEVATER Academic Track (test)
0.93CIFAR10 AccuracySwin-B
Evaluation Results
| Method | Links | ||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Swin-BEvaluation Protocol=Zero-shot, Pre-training Data=UniCL (ImageNet-21K + GCC + YFCC), Input Resolution=384x3842022.03 | 0.93 | 0.695 | 0.073 | 0.255 | 0.244 | 0.304 | 0.027 | 0.71 | 0.09 | 0.526 | 0.124 | 0.101 | 0.704 | 0.524 | 0.506 | 0.501 | 0.448 | 0.138 | 0.813 | 0.432 | |
| FocalNet-BEvaluation Protocol=Zero-shot, Pre-training Data=UniCL (ImageNet-21K + GCC + YFCC), Input Resolution=384x3842022.03 | 0.902 | 0.678 | 0.067 | 0.254 | 0.353 | 0.308 | 0.035 | 0.683 | 0.111 | 0.51 | 0.179 | 0.113 | 0.717 | 0.449 | 0.521 | 0.495 | 0.414 | 0.242 | 0.813 | 0.44 |