Regnety 640.seer

🚀 RegNetY-640 SEER模型卡

RegNetY-640 SEER是一個RegNetY-64GF特徵/骨幹模型。它根據SEER方法進行預訓練，即在“20億張隨機互聯網圖像”上使用SwAV進行自監督學習。

SEER採用SEER許可證，版權歸Meta Platforms, Inc.所有。該許可證為非商業許可證，對使用和分發有一定限制。

timm庫中的RegNet實現包含了許多其他實現中沒有的增強功能，包括：

• 隨機深度
• 梯度檢查點
• 逐層學習率衰減
• 可配置的輸出步長（膨脹率）
• 可配置的激活層和歸一化層
• 用於RegNetV變體的預激活瓶頸塊選項
• 唯一已知的帶有預訓練權重的RegNetZ模型定義

✨ 主要特性

• 模型類型：圖像分類/特徵骨幹
• 模型統計信息：
  • 參數數量（M）：276.5
  • GMACs：64.2
  • 激活值（M）：42.5
  • 圖像大小：224 x 224
• 相關論文：
  • Self-supervised Pretraining of Visual Features in the Wild: https://arxiv.org/abs/2103.01988v2
  • Designing Network Design Spaces: https://arxiv.org/abs/2003.13678
• 原始代碼庫：https://github.com/facebookresearch/vissl
• 預訓練數據集：RandomInternetImages-2B

💻 使用示例

基礎用法

圖像分類

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('regnety_640.seer', 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)

特徵圖提取

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(
    'regnety_640.seer',
    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, 32, 112, 112])
    #  torch.Size([1, 328, 56, 56])
    #  torch.Size([1, 984, 28, 28])
    #  torch.Size([1, 1968, 14, 14])
    #  torch.Size([1, 4920, 7, 7])

    print(o.shape)

圖像嵌入

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(
    'regnety_640.seer',
    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, 4920, 7, 7) shaped tensor

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

📚 詳細文檔

模型比較

你可以在timm 模型結果中探索該模型的數據集和運行時指標。

在以下比較總結中，標記為ra_in1k、ra3_in1k、ch_in1k、sw_*和lion_*的權重是在timm中訓練的。

📄 許可證

該模型採用SEER許可證，版權歸Meta Platforms, Inc.所有。該許可證為非商業許可證，對使用和分發有一定限制。

🔧 技術細節

該模型基於RegNetY架構，通過SEER方法在大規模隨機互聯網圖像上進行自監督預訓練。SEER方法利用SwAV技術，在無監督的情況下學習圖像特徵，使得模型在視覺任務中具有更好的魯棒性和公平性。

timm庫中的實現包含了許多增強功能，如隨機深度、梯度檢查點、逐層學習率衰減等，這些功能有助於提高模型的訓練效率和性能。

📝 引用

如果你在研究中使用了該模型，請引用以下論文：

@article{goyal2022vision,
  title={Vision Models Are More Robust And Fair When Pretrained On Uncurated Images Without Supervision}, 
  author={Priya Goyal and Quentin Duval and Isaac Seessel and Mathilde Caron and Ishan Misra and Levent Sagun and Armand Joulin and Piotr Bojanowski},
  year={2022},
  eprint={2202.08360},
  archivePrefix={arXiv},
  primaryClass={cs.CV}
}

@InProceedings{Radosavovic2020,
  title = {Designing Network Design Spaces},
  author = {Ilija Radosavovic and Raj Prateek Kosaraju and Ross Girshick and Kaiming He and Piotr Doll{'a}r},
  booktitle = {CVPR},
  year = {2020}
}

@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}}
}