| object-detection-on-coco-o | EVA | #1 | Average mAP: 57.8Effective Robustness: 28.86 |
| instance-segmentation-on-coco-test-dev | EVA | #3 | mask AP: 55.5AP50: 80.0APS: 36.3APM: 58.0APL: 72.4 |
| instance-segmentation-on-lvis-v1-0-val | EVA | #3 | mask AP: 55.0 |
| semantic-segmentation-on-ade20k-val | EVA | #3 | mIoU: 61.5 |
| semantic-segmentation-on-coco-stuff-test | EVA | #3 | mIoU: 53.4 |
| domain-generalization-on-imagenet-sketch | EVA | #4 | Top-1 accuracy: 67.7 |
| object-detection-on-lvis-v1-0-val | EVA | #4 | box AP: 62.2box APr: 55.1 |
| instance-segmentation-on-coco-minival | EVA | #5 | mask AP: 55.0AP50: 79.4AP75: 60.9APL: 72.0APM: 58.4APS: 37.6 |
| semantic-segmentation-on-ade20k | EVA | #6 | Validation mIoU: 62.3Params (M): 1074 |
| action-classification-on-kinetics-700 | EVA | #7 | Top-1 Accuracy: 82.9% |
| object-detection-on-coco-test-dev | EVA (Cascade Mask R-CNN, TTA) | #7 | box mAP: 64.7 |
| object-detection-on-coco-val2017 | EVA (Cascade Mask R-CNN, TTA) | #8 | box AP: 64.5 |
| image-classification-on-imagenet-v2 | EVA | #9 | Top 1 Accuracy: 81.6 |
| action-classification-on-kinetics-600 | EVA | #11 | Top-1 Accuracy: 89.8% |
| action-classification-on-kinetics-400 | EVA | #13 | Acc@1: 89.7 |
| image-classification-on-imagenet-real | EVA | #13 | Accuracy: 90.8% |
| image-classification-on-imagenet | EVA | #16 | Top 1 Accuracy: 89.7%Number of params: 1000M |