Swin Transformer - Image input size

[loukasilias]

Hi everyone,

I would like to ask you if the image used as input to the swin transformer will have input size [channels, H, W] or [H,W,channels].
Moreover, regarding the feature extraction, I would like to ask you if the code below is right.

model = timm.create_model('swin_base_patch4_window7_224', pretrained=True)
model.head = nn.Linear(1024,2)
from timm.data import resolve_data_config
from timm.data.transforms_factory import create_transform

config = resolve_data_config({}, model=model)
transform = create_transform(**config)
image = transform(image)

logits = model(image)