resnetv2_50x3_bit.goog_in21k_ft_in1k 的模型卡片

一个 ResNet-V2-BiT（带有预激活 ResNet 的大规模迁移）图像分类模型。由论文作者在 ImageNet-21k 上进行预训练，并在 ImageNet-1k 上进行微调。

此模型使用：

• 与批归一化（BN）相比，组归一化（GN）结合权重标准化（WS）进行训练。

模型细节

• 模型类型: 图像分类 / 特征骨干
• 模型统计信息:
  • 参数数目（百万）: 217.3
  • GMACs: 145.7
  • 激活数目（百万）: 133.4
  • 图像尺寸: 448 x 448
• 相关论文:
  • Big Transfer（BiT）: 通用视觉表示学习: https://arxiv.org/abs/1912.11370
  • 深度残差网络中的标识映射: https://arxiv.org/abs/1603.05027
• 数据集: ImageNet-1k
• 预训练数据集: ImageNet-21k
• 原始论文: https://github.com/google-research/big_transfer

模型用途

图像分类

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('resnetv2_50x3_bit.goog_in21k_ft_in1k', pretrained=True)
model = model.eval()

# get model specific transforms (normalization, resize)
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 single image into batch of 1

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

特征图提取

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(
    'resnetv2_50x3_bit.goog_in21k_ft_in1k',
    pretrained=True,
    features_only=True,
)
model = model.eval()

# get model specific transforms (normalization, resize)
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 single image into batch of 1

for o in output:
    # print shape of each feature map in output
    # e.g.:
    #  torch.Size([1, 192, 224, 224])
    #  torch.Size([1, 768, 112, 112])
    #  torch.Size([1, 1536, 56, 56])
    #  torch.Size([1, 3072, 28, 28])
    #  torch.Size([1, 6144, 14, 14])

    print(o.shape)

图像嵌入

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(
    'resnetv2_50x3_bit.goog_in21k_ft_in1k',
    pretrained=True,
    num_classes=0,  # remove classifier nn.Linear
)
model = model.eval()

# get model specific transforms (normalization, resize)
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))  # output is (batch_size, num_features) shaped tensor

# or equivalently (without needing to set num_classes=0)

output = model.forward_features(transforms(img).unsqueeze(0))
# output is unpooled, a (1, 6144, 14, 14) shaped tensor

output = model.forward_head(output, pre_logits=True)
# output is a (1, num_features) shaped tensor

模型比较

在 timm 中探索此模型的数据集和运行时指标 model results .

引用

@inproceedings{Kolesnikov2019BigT,
  title={Big Transfer (BiT): General Visual Representation Learning},
  author={Alexander Kolesnikov and Lucas Beyer and Xiaohua Zhai and Joan Puigcerver and Jessica Yung and Sylvain Gelly and Neil Houlsby},
  booktitle={European Conference on Computer Vision},
  year={2019}
}

@article{He2016,
  author = {Kaiming He and Xiangyu Zhang and Shaoqing Ren and Jian Sun},
  title = {Identity Mappings in Deep Residual Networks},
  journal = {arXiv preprint arXiv:1603.05027},
  year = {2016}
}

@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}}
}