timm/swin_large_patch4_window12_384.ms_in22k

Clasificación de Imágenes

from urllib.request import urlopen
from PIL import Image
import timm

img = Image.open(urlopen('https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/beignets-task-guide.png'))

model = timm.create_model('swin_large_patch4_window12_384.ms_in22k', pretrained=True)
model = model.eval()

# obten transformaciones específicas del modelo (normalización, redimensionado)
data_config = timm.data.resolve_model_data_config(model)
transforms = timm.data.create_transform(**data_config, is_training=False)

output = model(transforms(img).unsqueeze(0)) # agregar imagen individual en lote de 1

top5_probabilities, top5_class_indices = torch.topk(output.softmax(dim=1) * 100, k=5)

Extracción de Mapa de Características

from urllib.request import urlopen
from PIL import Image
import timm

img = Image.open(urlopen('https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/beignets-task-guide.png'))

model = timm.create_model('swin_large_patch4_window12_384.ms_in22k', pretrained=True, features_only=True)
model = model.eval()

# obten transformaciones específicas del modelo (normalización, redimensionado)
data_config = timm.data.resolve_model_data_config(model)
transforms = timm.data.create_transform(**data_config, is_training=False)

output = model(transforms(img).unsqueeze(0)) # agregar imagen individual en lote de 1

for o in output:
    print(o.shape)

Embeddings de Imágenes

from urllib.request import urlopen
from PIL import Image
import timm

img = Image.open(urlopen('https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/beignets-task-guide.png'))

model = timm.create_model('swin_large_patch4_window12_384.ms_in22k', pretrained=True, num_classes=0)
model = model.eval()

# obten transformaciones específicas del modelo (normalización, redimensionado)
data_config = timm.data.resolve_model_data_config(model)
transforms = timm.data.create_transform(**data_config, is_training=False)

output = model(transforms(img).unsqueeze(0)) # la salida es un tensor con forma (batch_size, num_features)

output = model.forward_features(transforms(img).unsqueeze(0))
output = model.forward_head(output, pre_logits=True)