EndoUC 溃疡性结肠炎内镜数据集

数据集获取

源站：https://synapse.org/Synapse:syn75978330

标准获取流程：

1. 访问源站链接，注册/登录账号；
2. 进入数据集详情页，按页面指引申请权限或直接下载；
3. 部分平台提供 SDK，例如：
   • OpenDataLab: pip install openxlab 后 openxlab.dataset.get(...)
   • ModelScope: pip install modelscope 后 modelscope download
   • 天池: 登录后页面提供直链下载
     以源站「数据下载/使用文档」页面为准。

数据说明

About EndoUC dataset Please note that redistribution is not allowed and any copies provided online should be deleted immediately. You are not allowed to post in Kaggle or Dataset Ninja or in your webpages but rather use Synapse site to download the dataset. Any misuse will be actioned!!! EndoUC is a multicenter gastrointestinal endoscopy dataset designed for the evaluation and assessment of Ulcerative Colitis (UC). It contains 2,495 endoscopic image frames collected from multiple clinical centers. The dataset is organized into three subsets, each representing a different medical center and/or imaging quality. Each image is accompanied by detailed clinical annotations, including the Mayo Endoscopic Score (MES) and the Ulcerative Colitis Endoscopic Index of Severity (UCEIS), as well as annotations for vascular pattern, bleeding, erosions and ulcers, friability, erythema, and a comprehensive ground-truth clinical caption (“description”). The Mayo Score and UCEIS annotations are suitable for endoscopic severity classification tasks, while the ground-truth clinical descriptions enable image captioning and vision-language modeling applications. Thanks to all collaborators. A part of this dataset was initially used for the 4th international workshop and challenge on Endoscopic computer vision. The challenge is designed to collaborate and curate multicenter datasets and promote building of generalisable models and assess deep learning methods. The provided dataset is an extended version of EndoUC2026 challenge. You must cite all four papers as these contain different collaboration works towards the dataset compilation. If the authors do not cite these works then they violate our data usage policy: [`1] Borgli, H., Thambawita, V., Smedsrud, P.H., Hicks, S., Jha, D., Eskeland, S.L., Randel, K.R., Pogorelov, K., Lux, M., Nguyen, D.T.D. and Johansen, D., 2020. HyperKvasir, a comprehensive multi-class image and video dataset for

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。