bat_resnext26ts.ch_in1k

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('bat_resnext26ts.ch_in1k', pretrained=True)
model = model.eval()

# obtener las transformaciones específicas del modelo (normalización, cambio de tamaño)
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)) # unsqueeze imagen única en un batch de 1

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

Extracción de Mapas 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(
'bat_resnext26ts.ch_in1k',
pretrained=True,
features_only=True,
)
model = model.eval()

# obtener las transformaciones específicas del modelo (normalización, cambio de tamaño)
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)) # unsqueeze imagen única en un batch de 1

for o in output:
# imprimir la forma de cada mapa de características en la salida
# ej.:
#  torch.Size([1, 64, 128, 128])
#  torch.Size([1, 256, 64, 64])
#  torch.Size([1, 512, 32, 32])
#  torch.Size([1, 1024, 16, 16])
#  torch.Size([1, 2048, 8, 8])

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(
'bat_resnext26ts.ch_in1k',
pretrained=True,
num_classes=0, # eliminar el clasificador nn.Linear
)
model = model.eval()

# obtener las transformaciones específicas del modelo (normalización, cambio de tamaño)
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)

# o equivalentemente (sin necesidad de establecer num_classes=0)

output = model.forward_features(transforms(img).unsqueeze(0))
# la salida no está agrupada, un tensor con forma (1, 2048, 8, 8)

output = model.forward_head(output, pre_logits=True)
# la salida es un tensor con forma (1, num_features)

Citación

@misc{rw2019timm,
author = {Ross Wightman},
title = {PyTorch Image Models},
year = {2019},
publisher = {GitHub},
journal = {GitHub repository},
doi = {10.5281/zenodo.4414861},
howpublished = {\url{https://github.com/huggingface/pytorch-image-models}}
}

@InProceedings{Chi_2020_CVPR,
author = {Chi, Lu and Yuan, Zehuan and Mu, Yadong and Wang, Changhu},
title = {Non-Local Neural Networks With Grouped Bilinear Attentional Transforms},
booktitle = {The IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)},
month = {June},
year = {2020}
}

@article{Xie2016,
title={Aggregated Residual Transformations for Deep Neural Networks},
author={Saining Xie and Ross Girshick and Piotr Dollár and Zhuowen Tu and Kaiming He},
journal={arXiv preprint arXiv:1611.05431},
year={2016}
}