简化版Swin Transformer模型卡片：swinv2_cr_tiny_ns_224.sw_in1k

这是在官方代码发布前独立实现的Swin Transformer V2模型。由 Christoph Reich 和 Ross Wightman 共同合作完成，该模型与官方实现有一些差异:

• MLP的相对位置偏置采用非标准化的自然对数，而不是标准化的Sigmoid函数，经过缩放后取以2为底的对数。
• 在每个Stage的末尾添加了LayerNorm的选项（"ns"变体）。
• 在每个Stage和最终特征输出处，默认使用NCHW张量布局。
• 由 Ross Wightman 在ImageNet-1K上进行预训练。

模型细节

• 模型类型：图像分类 / 特征骨干
• 模型统计数据：
  • 参数数量 (M): 28.3
  • GMACs: 4.7
  • 激活值数量 (M): 28.5
  • 图像尺寸: 224 x 224
• 相关论文：
  • Swin Transformer V2: 扩展容量和分辨率: https://arxiv.org/abs/2111.09883
• 原始代码： https://github.com/ChristophReich1996/Swin-Transformer-V2
• 数据集：ImageNet-1K

模型使用

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('swinv2_cr_tiny_ns_224.sw_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(
    'swinv2_cr_tiny_ns_224.sw_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. for swin_base_patch4_window7_224 (NHWC output)
    #  torch.Size([1, 56, 56, 128])
    #  torch.Size([1, 28, 28, 256])
    #  torch.Size([1, 14, 14, 512])
    #  torch.Size([1, 7, 7, 1024])
    # e.g. for swinv2_cr_small_ns_224 (NCHW output)
    #  torch.Size([1, 96, 56, 56]) 
    #  torch.Size([1, 192, 28, 28])
    #  torch.Size([1, 384, 14, 14])
    #  torch.Size([1, 768, 7, 7])
    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(
    'swinv2_cr_tiny_ns_224.sw_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 (ie.e a (batch_size, H, W,  num_features) tensor for swin / swinv2
# or (batch_size, num_features, H, W) for swinv2_cr

output = model.forward_head(output, pre_logits=True)
# output is (batch_size, num_features) tensor

模型比较

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

引用

@inproceedings{liu2021swinv2,
  title={Swin Transformer V2: Scaling Up Capacity and Resolution}, 
  author={Ze Liu and Han Hu and Yutong Lin and Zhuliang Yao and Zhenda Xie and Yixuan Wei and Jia Ning and Yue Cao and Zheng Zhang and Li Dong and Furu Wei and Baining Guo},
  booktitle={International Conference on Computer Vision and Pattern Recognition (CVPR)},
  year={2022}
}

@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}}
}