license: apache-2.0 library_name: timm tags:

• 图像特征提取
• timm
• transformers

vit_base_patch16_224.dino 模型卡

基于自监督DINO方法训练的Vision Transformer（ViT）图像特征模型。

模型详情

• 模型类型: 图像分类/特征主干网络
• 模型统计:
  • 参数量（百万）: 85.8
  • GMACs计算量: 16.9
  • 激活值（百万）: 16.5
  • 图像尺寸: 224 x 224
• 相关论文:
  • 《自监督视觉Transformer的新兴特性》: https://arxiv.org/abs/2104.14294
  • 《一张图像等于16x16个单词：大规模图像识别的Transformer》: https://arxiv.org/abs/2010.11929v2
• 预训练数据集: ImageNet-1k
• 原始实现: https://github.com/facebookresearch/dino

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('vit_base_patch16_224.dino', pretrained=True)
model = model.eval()

# 获取模型特定的变换（归一化、调整大小）
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))  # 将单张图像扩展为批量大小为1

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

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(
    'vit_base_patch16_224.dino',
    pretrained=True,
    num_classes=0,  # 移除分类器nn.Linear
)
model = model.eval()

# 获取模型特定的变换（归一化、调整大小）
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))  # 输出为（批量大小, 特征数量）形状的张量

# 或等效方式（无需设置num_classes=0）

output = model.forward_features(transforms(img).unsqueeze(0))
# 输出未池化，为（1, 197, 768）形状的张量

output = model.forward_head(output, pre_logits=True)
# 输出为（1, 特征数量）形状的张量

模型比较

在timm的模型结果中探索该模型的数据集和运行时指标。

引用

@inproceedings{caron2021emerging,
  title={自监督视觉Transformer的新兴特性},
  author={Caron, Mathilde and Touvron, Hugo and Misra, Ishan and J{'e}gou, Herv{'e} and Mairal, Julien and Bojanowski, Piotr and Joulin, Armand},
  booktitle={IEEE/CVF国际计算机视觉会议论文集},
  pages={9650--9660},
  year={2021}
}

@article{dosovitskiy2020vit,
  title={一张图像等于16x16个单词：大规模图像识别的Transformer},
  author={Dosovitskiy, Alexey and Beyer, Lucas and Kolesnikov, Alexander and Weissenborn, Dirk and Zhai, Xiaohua and Unterthiner, Thomas and  Dehghani, Mostafa and Minderer, Matthias and Heigold, Georg and Gelly, Sylvain and Uszkoreit, Jakob and Houlsby, Neil},
  journal={ICLR},
  year={2021}
}

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