MaxViT-Tiny backbone (Tu et al., 2022).
Builds the canonical MaxViT-Tiny configuration:
depths=(2, 2, 5, 2), dims=(64, 128, 256, 512),
window_size=7. Approximately 31M parameters.
Model Size
Parameters
pretrainedbool= FalseIf
True, loads ImageNet-1k pretrained weights when
available in the model zoo. Defaults to False.**overridesobject= {}Keyword overrides applied on top of the canonical MaxViT-Tiny
config. Each override must match a field of
MaxViTConfig.Returns
MaxViTA MaxViT backbone returning a flat
feature.
Notes
MaxViT-Tiny reaches 83.6% top-1 on ImageNet-1k at 224x224 (Tu et al., 2022, Table 6). See arXiv:2204.01697.
Examples
>>> import lucid
>>> from lucid.models.vision.maxvit import maxvit_tiny
>>> model = maxvit_tiny()
>>> x = lucid.randn(1, 3, 224, 224)
>>> feat = model.forward_features(x)
>>> feat.shape
(1, 512)