towhee.models.omnivore.omnivore.omnivore_swint

towhee.models.omnivore.omnivore.omnivore_swint(pretrained: bool = True, progress: bool = True, load_heads: bool = True, **kwargs: Any)torch.nn.modules.module.Module[source]

Omnivore model trunk: Swin T patch (2,4,4) window (8,7,7) :param pretrained: if True loads weights from model trained on

Imagenet 1k, Kinetics 400, SUN RGBD.

Parameters:
  • progress – print progress of loading checkpoint

  • load_heads – if True, loads the 3 heads, one each for image/video/rgbd prediction. If False loads only the trunk.

Returns:

nn.Module of the omnivore model

Return type:

model