towhee.models.omnivore.omnivore¶

Functions

`create_model`
`get_all_heads`
`get_imagenet_head`
`get_kinetics_head`
`get_sunrgbd_head`
`omnivore_swinb`	Omnivore model trunk: Swin B patch (2,4,4) window (1,6,7,7) :param pretrained: if True loads weights from model trained on Imagenet 1k, Kinetics 400, SUN RGBD. :param progress: print progress of loading checkpoint :param load_heads: if True, loads the 3 heads, one each for image/video/rgbd prediction. If False loads only the trunk.
`omnivore_swinb_epic`	Omnivore swin B model trained on EPIC-KITCHENS-100 dataset :param progress: print progress of loading checkpoint
`omnivore_swinb_imagenet21k`	Omnivore Swin B model pretrained on Imagenet 1k, Imagenet 21k, Kinetics 400, SUN RGBD. By default the pretrained weights will be loaded. :param progress: print progress of loading checkpoint :param load_heads: if True, loads the 3 heads, one each for image/video/rgbd prediction. If False loads only the trunk.
`omnivore_swinl_imagenet21k`	Swin L patch 244 window 877 pretrained on Imagenet 1k, Imagenet 21k, Kinetics 400, SUN RGBD. By default the pretrained weights will be loaded. :param pretrained: if True loads weights from model trained on Imagenet1k, Imagenet 21k, Kinetics 400, SUN RGBD. :param progress: print progress of loading checkpoint :param load_heads: if True, loads the 3 heads, one each for image/video/rgbd prediction. If False loads only the trunk.
`omnivore_swinl_kinetics600`	Swin L patch 244 window 877 trained with Kinetics 600 :param pretrained: if True loads weights from model trained on Imagenet 1k, Kinetics 600, SUN RGBD. :param progress: print progress of loading checkpoint :param load_heads: if True, loads the 3 heads, one each for image/video/rgbd prediction. If False loads only the trunk.
`omnivore_swins`	Omnivore model trunk: Swin S patch (2,4,4) window (8,7,7) :param pretrained: if True loads weights from model trained on Imagenet 1k, Kinetics 400, SUN RGBD. :param progress: print progress of loading checkpoint :param load_heads: if True, loads the 3 heads, one each for image/video/rgbd prediction. If False loads only the trunk.
`omnivore_swint`	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. :param progress: print progress of loading checkpoint :param load_heads: if True, loads the 3 heads, one each for image/video/rgbd prediction. If False loads only the trunk.

Classes

OmnivoreModel

Args: