1 paper
Tobias Christian Nauen, Anosh Billimoria, Federico Raue +3
Comparing transformer backbones for image segmentation is confounded: each is paired with a different decoder, recipe, and pretraining, so reported differences rarely reflect the b…