eca_nfnet_l1.ra2_in1k 模型卡片

ECA-NFNet-Lite （带 ECA 注意力的轻量级 NFNet）图像分类模型。由 Ross Wightman 在 timm 中训练。

无标准化网络是类似于（预激活）ResNet的模型，没有任何标准化层。它们使用缩放权重标准化和基于信号传播分析在残差路径和非线性处具体放置标量增益，而不是批标准化或其他替代方法。

轻量级 NFNets 是 timm 具体的变种，它们将 SE 和 bottleneck 比例从 0.5 -> 0.25（减小宽度）并使用较小的 group 大小，同时保持相同的深度。使用 SiLU 激活函数替代 GELU。

此 NFNet 变种还使用了 ECA（高效通道注意力），而不是 SE（Squeeze-and-Excitation）。

模型详情

• 模型类型：图像分类 / 特征主干
• 模型统计信息：
  • 参数数目（M）：41.4
  • GMACs（十亿次浮点数乘法-加法运算）：9.6
  • 激活数目（M）：22.0
  • 图像尺寸：训练集 = 256 x 256，测试集 = 320 x 320
• 论文：
  • 无标准化的高性能大规模图像识别： https://arxiv.org/abs/2102.06171
  • 补偿非标准化 ResNet 中的性能差距的信号传播特性： https://arxiv.org/abs/2101.08692
• 原始文献： https://github.com/huggingface/pytorch-image-models
• 数据集：ImageNet-1k

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('eca_nfnet_l1.ra2_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(
    'eca_nfnet_l1.ra2_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, 64, 128, 128])
    #  torch.Size([1, 256, 64, 64])
    #  torch.Size([1, 512, 32, 32])
    #  torch.Size([1, 1536, 16, 16])
    #  torch.Size([1, 3072, 8, 8])

    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(
    'eca_nfnet_l1.ra2_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, 3072, 8, 8) shaped tensor

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

模型比较

在 timm 中探索该模型的数据集和运行时指标： model results

引用

@article{brock2021high,
  author={Andrew Brock and Soham De and Samuel L. Smith and Karen Simonyan},
  title={High-Performance Large-Scale Image Recognition Without Normalization},
  journal={arXiv preprint arXiv:2102.06171},
  year={2021}
}

@inproceedings{brock2021characterizing,
  author={Andrew Brock and Soham De and Samuel L. Smith},
  title={Characterizing signal propagation to close the performance gap in
  unnormalized ResNets},
  booktitle={9th International Conference on Learning Representations, {ICLR}},
  year={2021}
}

@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}}
}