fcn_resnet101(pretrained: bool | str = False, weights: FCNResNet101Weights | None = None, overrides: object = {})FCN with ResNet-101 backbone (Long et al., CVPR 2015).
Builds an FCNForSemanticSegmentation with the same head
configuration as fcn_resnet50 but a deeper dilated ResNet-101
backbone (23 blocks in layer3 vs 6 for ResNet-50). Typically gains
1-2 mIoU points on Pascal VOC at the cost of ~70% more FLOPs.
Model Size
Parameters
pretrainedbool or str= FalsePretrained-weight selector.
False → random init; True →
the DEFAULT tag (FCNResNet101Weights.COCO_WITH_VOC_LABELS_V1);
a tag string → that specific checkpoint. Mutually exclusive with
weights (which wins if both are given).Explicit weights enum member. Takes precedence over
pretrained.**overridesobject= {}Keyword overrides forwarded into
FCNConfig.Returns
FCNForSemanticSegmentationSegmentation model with the FCN-R101 configuration applied (or
with overrides merged on top of it).
Examples
>>> import lucid
>>> from lucid.models.vision.fcn import fcn_resnet101
>>> model = fcn_resnet101()
>>> x = lucid.randn(1, 3, 512, 512)
>>> out = model(x)
>>> out.logits.shape
(1, 21, 512, 512)