数据集获取

git clone https://github.com/v7labs/covid-19-xray-dataset.git

curl -L -o repo.zip https://github.com/v7labs/covid-19-xray-dataset/archive/refs/heads/master.zip
unzip repo.zip

源站 README 摘录（使用方式）

COVID-19 xray dataset

BROWSE & DOWNLOAD THE DATASET ON V7 DARWIN HERE
or run the following command from Darwin-py to download the latest version

pip install darwin-py
darwin dataset pull v7-labs/covid-19-chest-x-ray-dataset:all-images

This dataset contains 6500 images of AP/PA chest x-rays with pixel-level polygonal lung segmentations. There are 517 cases of COVID-19 amongst these.
Use the command below to download only images presenting COVID-19.

darwin dataset pull v7-labs/covid-19-chest-x-ray-dataset:covid-only

WARNING: This dataset is not intended for use in clinical diagnostics.
Each image contains:

• Two “Lung” segmentation masks (rendered as polygons, including the posterior region behind the heart).
• A tag for the type of pneumonia (viral, bacterial, fungal, healthy/none)
• If the patient has COVID-19, additional tags stating age, sex, temperature, location, intubation status, ICU admission, and patient outcome.
  Lung annotations are polygons following pixel-level boundaries. These can be exported as COCO, VOC, or Darwin JSON formats. Each annotation file contains a URL to the original full resolution image, as well as a reduced size thumbnail.

LUNG SEGMENTATION NOTES: Lung segmentations in this dataset include most of the heart, revealing lung opacities behind the heart which may be relevant for assessing the severity of viral pneumonia. Uniformly shaped lungs also de-couples the shape and content within the left lung from the size of the heart.

The lower-most part of the lungs is defined by the extent of the diaphragm, where visible. If the back of the lungs is clearly visible through the diaphragm it is also included up until the lower-most visible part of the lungs.

Lung segmentations were performed by human annotators using Auto-Annotate, adjusted, and reviewed by humans.

Other important notes:

• Image resolutions, sources, and orientations vary across the dataset, with the largest image being 5600x4700 and smallest being 156x156. You may sort images by dimensions on [Darwin](https://darwin

数据加载示例（图像类）

from PIL import Image
import glob, os

files = (glob.glob(os.path.join(path, "**", "*.png"), recursive=True)
       + glob.glob(os.path.join(path, "**", "*.jpg"), recursive=True)
       + glob.glob(os.path.join(path, "**", "*.tif"), recursive=True))
print("图像文件数:", len(files))
img = Image.open(files[0]); print("尺寸/模式:", img.size, img.mode)

# torchvision Dataset 方式:
# from torchvision import datasets
# ds = datasets.ImageFolder(path)  # 要求 子目录=类别

目录组织与标注格式以源站说明和下载后实际文件为准。

完整仓库：github.com/v7labs/covid-19-xray-dataset