SERE: Exploring Feature Self-relation for Self-supervised Transformer

Open paper
Benchmark Model Rank Results
semantic-segmentation-on-imagenet-sSERE (ViT-B/16, 100ep, 224x224, SSL+FT)#2mIoU (val): 63.0mIoU (test): 63.3
semantic-segmentation-on-imagenet-sSERE (ViT-S/16, 100ep, 224x224, SSL+FT, mmseg)#6mIoU (val): 59.4mIoU (test): 59.0
semantic-segmentation-on-imagenet-sSERE (ViT-S/16, 100ep, 224x224, SSL+FT)#7mIoU (val): 58.9mIoU (test): 57.8
semantic-segmentation-on-imagenet-sSERE (ViT-B/16, 100ep, 224x224, SSL)#12mIoU (val): 48.6mIoU (test): 48.2
semantic-segmentation-on-imagenet-sSERE (ViT-S/16, 100ep, 224x224, SSL, mmseg)#15mIoU (val): 41.0mIoU (test): 40.5
semantic-segmentation-on-imagenet-sSERE (ViT-S/16, 100ep, 224x224, SSL)#16mIoU (val): 41.0mIoU (test): 40.2

Rank counts only results with a code link.