class
AutoModelForSemanticSegmentation
extends
_BaseAutoClassAutoModelForSemanticSegmentation()Auto-dispatching wrapper that loads a semantic-segmentation model.
Resolves names like "fcn_resnet50" / "unet" /
"maskformer_resnet50" to their concrete
{Family}ForSemanticSegmentation subclass.
Notes
The forward returns a SemanticSegmentationOutput whose
logits carry per-pixel class scores of shape
(B, num_classes, H, W).
Examples
>>> from lucid.models import AutoModelForSemanticSegmentation
>>> model = AutoModelForSemanticSegmentation.from_pretrained("fcn_resnet50")
>>> type(model).__name__
'FCNForSemanticSegmentation'