regnet_y_16gf

torchvision.models.regnet_y_16gf(*, weights: Optional[RegNet_Y_16GF_Weights] = None, progress: bool = True, **kwargs: Any) → RegNet

Constructs a RegNetY_16GF architecture from Designing Network Design Spaces.

Parameters:

• weights (RegNet_Y_16GF_Weights, optional) – The pretrained weights to use. See RegNet_Y_16GF_Weights below for more details and possible values. By default, no pretrained weights are used.

• progress (bool, optional) – If True, displays a progress bar of the download to stderr. Default is True.

• **kwargs – parameters passed to either torchvision.models.regnet.RegNet or torchvision.models.regnet.BlockParams class. Please refer to the source code for more detail about the classes.

class torchvision.models.RegNet_Y_16GF_Weights(value)

The model builder above accepts the following values as the weights parameter. RegNet_Y_16GF_Weights.DEFAULT is equivalent to RegNet_Y_16GF_Weights.IMAGENET1K_V2. You can also use strings, e.g. weights='DEFAULT' or weights='IMAGENET1K_V1'.

RegNet_Y_16GF_Weights.IMAGENET1K_V1:

These weights reproduce closely the results of the paper using a simple training recipe.

acc@1 (on ImageNet-1K)	80.424
acc@5 (on ImageNet-1K)	95.24
min_size	height=1, width=1
categories	tench, goldfish, great white shark, … (997 omitted)
num_params	83590140
recipe	link

The inference transforms are available at RegNet_Y_16GF_Weights.IMAGENET1K_V1.transforms and perform the following preprocessing operations: Accepts PIL.Image, batched (B, C, H, W) and single (C, H, W) image torch.Tensor objects. The images are resized to resize_size=[256] using interpolation=InterpolationMode.BILINEAR, followed by a central crop of crop_size=[224]. Finally the values are first rescaled to [0.0, 1.0] and then normalized using mean=[0.485, 0.456, 0.406] and std=[0.229, 0.224, 0.225].

RegNet_Y_16GF_Weights.IMAGENET1K_V2:

These weights improve upon the results of the original paper by using a modified version of TorchVision’s new training recipe. Also available as RegNet_Y_16GF_Weights.DEFAULT.

acc@1 (on ImageNet-1K)	82.886
acc@5 (on ImageNet-1K)	96.328
min_size	height=1, width=1
categories	tench, goldfish, great white shark, … (997 omitted)
num_params	83590140
recipe	link

The inference transforms are available at RegNet_Y_16GF_Weights.IMAGENET1K_V2.transforms and perform the following preprocessing operations: Accepts PIL.Image, batched (B, C, H, W) and single (C, H, W) image torch.Tensor objects. The images are resized to resize_size=[232] using interpolation=InterpolationMode.BILINEAR, followed by a central crop of crop_size=[224]. Finally the values are first rescaled to [0.0, 1.0] and then normalized using mean=[0.485, 0.456, 0.406] and std=[0.229, 0.224, 0.225].

RegNet_Y_16GF_Weights.IMAGENET1K_SWAG_E2E_V1:

These weights are learnt via transfer learning by end-to-end fine-tuning the original SWAG weights on ImageNet-1K data.

acc@1 (on ImageNet-1K)	86.012
acc@5 (on ImageNet-1K)	98.054
min_size	height=1, width=1
categories	tench, goldfish, great white shark, … (997 omitted)
recipe	link
license	link
num_params	83590140

The inference transforms are available at RegNet_Y_16GF_Weights.IMAGENET1K_SWAG_E2E_V1.transforms and perform the following preprocessing operations: Accepts PIL.Image, batched (B, C, H, W) and single (C, H, W) image torch.Tensor objects. The images are resized to resize_size=[384] using interpolation=InterpolationMode.BICUBIC, followed by a central crop of crop_size=[384]. Finally the values are first rescaled to [0.0, 1.0] and then normalized using mean=[0.485, 0.456, 0.406] and std=[0.229, 0.224, 0.225].

RegNet_Y_16GF_Weights.IMAGENET1K_SWAG_LINEAR_V1:

These weights are composed of the original frozen SWAG trunk weights and a linear classifier learnt on top of them trained on ImageNet-1K data.

acc@1 (on ImageNet-1K)	83.976
acc@5 (on ImageNet-1K)	97.244
min_size	height=1, width=1
categories	tench, goldfish, great white shark, … (997 omitted)
recipe	link
license	link
num_params	83590140

The inference transforms are available at RegNet_Y_16GF_Weights.IMAGENET1K_SWAG_LINEAR_V1.transforms and perform the following preprocessing operations: Accepts PIL.Image, batched (B, C, H, W) and single (C, H, W) image torch.Tensor objects. The images are resized to resize_size=[224] using interpolation=InterpolationMode.BICUBIC, followed by a central crop of crop_size=[224]. Finally the values are first rescaled to [0.0, 1.0] and then normalized using mean=[0.485, 0.456, 0.406] and std=[0.229, 0.224, 0.225].