Pretrained weights for lucid.models.vgg_13_bn_cls.
VGG-B with BatchNorm (133.1 M params, 11.3 GFLOPs, top-1 71.59%).
Attributes
IMAGENET1K_V1WeightEntryImageNet-1k V1 checkpoint (top-1 71.586% / top-5 90.374%),
sourced from
torchvision/VGG13_BN_Weights.IMAGENET1K_V1.DEFAULTWeightEntryAlias for
IMAGENET1K_V1.Notes
Reference: Simonyan, Zisserman, "Very Deep Convolutional Networks for Large-Scale Image Recognition", ICLR 2015 (arXiv:1409.1556). BatchNorm added in the torchvision reimplementation (not in the original paper).
Examples
>>> from lucid.models import vgg_13_bn_cls
>>> model = vgg_13_bn_cls(pretrained=True).eval()