detectors.models package

Submodules

detectors.models.densenet module

Densenet models for CIFAR10, CIFAR100 and SVHN datasets.

detectors.models.densenet.densenet121_camelyon17(pretrained=False, **kwargs)

detectors.models.densenet.densenet121_cifar10(pretrained=False, **kwargs)

detectors.models.densenet.densenet121_cifar100(pretrained=False, **kwargs)

detectors.models.densenet.densenet121_svhn(pretrained=False, **kwargs)

detectors.models.resnet module

ResNet models for CIFAR10, CIFAR100, and SVHN datasets.

detectors.models.resnet.resnet18_cifar10(pretrained=False, **kwargs)

detectors.models.resnet.resnet18_cifar100(pretrained=False, **kwargs)

detectors.models.resnet.resnet18_svhn(pretrained=False, **kwargs)

detectors.models.resnet.resnet34_cifar10(pretrained=False, **kwargs)

detectors.models.resnet.resnet34_cifar100(pretrained=False, **kwargs)

detectors.models.resnet.resnet34_simclr_cifar10(pretrained=False, **kwargs)

detectors.models.resnet.resnet34_simclr_cifar100(pretrained=False, **kwargs)

detectors.models.resnet.resnet34_supcon_cifar10(pretrained=False, **kwargs)

detectors.models.resnet.resnet34_supcon_cifar100(pretrained=False, **kwargs)

detectors.models.resnet.resnet34_svhn(pretrained=False, **kwargs)

detectors.models.resnet.resnet50_cifar10(pretrained=False, **kwargs)

detectors.models.resnet.resnet50_cifar100(pretrained=False, **kwargs)

detectors.models.resnet.resnet50_simclr_cifar10(pretrained=False, **kwargs)

detectors.models.resnet.resnet50_simclr_cifar100(pretrained=False, **kwargs)

detectors.models.resnet.resnet50_supcon_cifar10(pretrained=False, **kwargs)

detectors.models.resnet.resnet50_supcon_cifar100(pretrained=False, **kwargs)

detectors.models.resnet.resnet50_svhn(pretrained=False, **kwargs)

detectors.models.utils module

class detectors.models.utils.ModelDefaultConfig(url: str, num_classes: int, input_size: Tuple[int, int, int] | Tuple[int, int] | int, pool_size: Tuple[int, int] | None, crop_pct: float, mean: Tuple[float, float, float], std: Tuple[float, float, float], first_conv: str, classifier: str, architecture: str, interpolation: str = 'bilinear', fixed_input_size: bool | None = False, **kwargs)

Bases: dict

Default configuration for models from timm library.

Example:

{
    'url': 'https://download.pytorch.org/models/resnet18-5c106cde.pth',
    'num_classes': 1000,
    'input_size': (3, 224, 224),
    'pool_size': (7, 7),
    'crop_pct': 0.875,
    'interpolation': 'bilinear',
    'fixed_input_size': True,
    'mean': (0.485, 0.456, 0.406),
    'std': (0.229, 0.224, 0.225),
    'first_conv': 'conv1',
    'classifier': 'fc',
    'architecture': 'resnet18'
}

detectors.models.utils.hf_hub_url_template(model_name: str)

detectors.models.vgg module

VGG models for CIFAR10, CIFAR100 and SVHN datasets.

detectors.models.vgg.vgg16_bn_cifar10(pretrained=False, **kwargs)

detectors.models.vgg.vgg16_bn_cifar100(pretrained=False, **kwargs)

detectors.models.vgg.vgg16_bn_svhn(pretrained=False, **kwargs)

detectors.models.vit module

Finetuned ViT models for CIFAR10, CIFAR100, and SVHN datasets.

detectors.models.vit.vit_base_patch16_224_in21k_ft_cifar10(pretrained=False, **kwargs)

detectors.models.vit.vit_base_patch16_224_in21k_ft_cifar100(pretrained=False, **kwargs)

detectors.models.vit.vit_base_patch16_224_in21k_ft_svhn(pretrained=False, **kwargs)

Module contents

detectors.models.create_transform(model: Module, is_training: bool = False)

Create a input transformation for a given model.

Based on the default configuration of the model following timm’s library.

Parameters:

• model (torch.nn.Module) – Model to create the transformation for.

• is_training (bool, optional) – Whether the transformation is for training or not. Defaults to False.

Returns:

The transformation.

Return type:

Callable