Model card for regetx032.tv2i1k

A RegNetX-3.2GF image classificatio model. Pretraied o ImageNet-1k by torchvisio cotributors (see ImageNet1K-V2 weight details https://github.com/pytorch/visio/issues/3995#ew-recipe).

The timm RegNet implemetatio icludes a umber of ehacemets ot preset i other implemetatios, icludig:

• stochastic depth
• gradiet checkpoitig
• layer-wise LR decay
• cofigurable output stride (dilatio)
• cofigurable activatio ad orm layers
• optio for a pre-activatio bottleeck block used i RegNetV variat
• oly kow RegNetZ model defiitios with pretraied weights

Model Details

• Model Type: Image classificatio / feature backboe
• Model Stats:
• Params (M): 15.3
• GMACs: 3.2
• Activatios (M): 11.4
• Image size: 224 x 224
• Papers:
• Desigig Network Desig Spaces: https://arxiv.org/abs/2003.13678
• Origial: https://github.com/pytorch/visio

Model Usage

Image Classificatio

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

img = Image.ope(urlope(
    'https://huggigface.co/datasets/huggigface/documetatio-images/resolve/mai/beigets-task-guide.pg'
))

model = timm.create_model('regetx_032.tv2_i1k', pretraied=True)
model = model.eval()

# get model specific trasforms (ormalizatio, resize)
data_cofig = timm.data.resolve_model_data_cofig(model)
trasforms = timm.data.create_trasform(**data_cofig, is_traiig=False)

output = model(trasforms(img).usqueeze(0))  # usqueeze sigle image ito batch of 1

top5_probabilities, top5_class_idices = torch.topk(output.softmax(dim=1) * 100, k=5)

Feature Map Extractio

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

img = Image.ope(urlope(
    'https://huggigface.co/datasets/huggigface/documetatio-images/resolve/mai/beigets-task-guide.pg'
))

model = timm.create_model(
    'regetx_032.tv2_i1k',
    pretraied=True,
    features_oly=True,
)
model = model.eval()

# get model specific trasforms (ormalizatio, resize)
data_cofig = timm.data.resolve_model_data_cofig(model)
trasforms = timm.data.create_trasform(**data_cofig, is_traiig=False)

output = model(trasforms(img).usqueeze(0))  # usqueeze sigle image ito batch of 1

for o i output:
    # prit shape of each feature map i output
    # e.g.:
    #  torch.Size([1, 32, 112, 112])
    #  torch.Size([1, 96, 56, 56])
    #  torch.Size([1, 192, 28, 28])
    #  torch.Size([1, 432, 14, 14])
    #  torch.Size([1, 1008, 7, 7])

    prit(o.shape)

Image Embeddigs

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

img = Image.ope(urlope(
    'https://huggigface.co/datasets/huggigface/documetatio-images/resolve/mai/beigets-task-guide.pg'
))

model = timm.create_model(
    'regetx_032.tv2_i1k',
    pretraied=True,
    um_classes=0,  # remove classifier .Liear
)
model = model.eval()

# get model specific trasforms (ormalizatio, resize)
data_cofig = timm.data.resolve_model_data_cofig(model)
trasforms = timm.data.create_trasform(**data_cofig, is_traiig=False)

output = model(trasforms(img).usqueeze(0))  # output is (batch_size, um_features) shaped tesor

# or equivaletly (without eedig to set um_classes=0)

output = model.forward_features(trasforms(img).usqueeze(0))
# output is upooled, a (1, 1008, 7, 7) shaped tesor

output = model.forward_head(output, pre_logits=True)
# output is a (1, um_features) shaped tesor

Model Compariso

Explore the dataset ad rutime metrics of this model i timm model results.

For the compariso summary below, the rai1k, ra3i1k, chi1k, sw, ad lio_ tagged weights are traied i timm.

Citatio

@IProceedigs{Radosavovic2020,
  title = {Desigig Network Desig Spaces},
  author = {Ilija Radosavovic ad Raj Prateek Kosaraju ad Ross Girshick ad Kaimig He ad Piotr Doll{'a}r},
  booktitle = {CVPR},
  year = {2020}
}

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

Model card for regnetx032.tv2in1k A RegNetX-3.2GF image classification model. Pretrained on ImageNet