ProstaTD
大规模多源机器人前列腺切除术视频数据集，含6万+标注帧和16万+手术三元组实例，用于手术器械、动作和目标检测。
基本信息
资源简介
ProstaTD
【数据集背景】
本数据集源站标题为《chen-yiliang/ProstaTD》。源站首次发布：2025-05-28。
【数据内容】
源站原始描述：ProstaTD: Bridging Surgical Triplet from Classification to Fully Supervised Detection ProstaTD is the first fully supervised surgical triplet detection dataset curated from 21 robot-assisted prostatectomy videos, collectively spanning full surgical procedures across multiple institutions, featuring 71,775 annotated frames with 196,490 structured surgical triplet instances (instrument-verb-target) that provide precise bounding box localization for all instruments alongside clinically validated te…
【适用方向】
影像类数据可用于医学图像分析模型（检测、分割、分类）的训练与评测，也可用于标注质量与成像参数相关性的研究。
【获取与许可】
源站页面：https://github.com/chen-yiliang/ProstaTD
使用许可：NOASSERTION。
【使用注意】
· 源站最后修改于 2026-05-31，引用时请注意版本时效。
下载信息
注册下载
需要注册 Kaggle 账号并登录后下载，适合需要跟踪下载记录和使用 API 的用户。
暂未开放公开下载
无需注册即可直接获取公开样本或文档，适合快速预览和评估数据集质量。
免登录有偿下载
公开数据集受托下载与技术交付服务。
提供高速下载与技术交付服务（收技术服务费，非数据销售）
暂未开放当前数据集主要来源为 Kaggle 公开托管，完整影像包建议通过原始链接或 Kaggle API 获取。
使用方式
数据集获取
git clone https://github.com/SmartHealthX/ProstaTD.git
curl -L -o repo.zip https://github.com/SmartHealthX/ProstaTD/archive/refs/heads/main.zip
unzip repo.zip
源站 README 摘录（使用方式）
📥 Download Access
To request access to the ProstaTD Dataset, please fill out our request form. Advice use Gmail for submission (other emails may find the reply in spam folders).
[!IMPORTANT]
There are two versions available:
• Version 1.1 (ProstaTDv1.1): includes only minimal refinements, such as removing a few overlapping instances. These marginal adjustments are not expected to affect model performance, and the experimental results should remain consistent with those reported in the paper.
• Version 2.0 (ProstaTDv2.0): is the refined version with corrections performed on the 20 fps frames. In v1, annotations were conducted solely using 1 fps image frames. However, the lack of temporal context at 1 fps often makes it difficult even for experienced surgeons to accurately identify surgical triplets. To address this, we expanded the dataset to 20 fps and corrected the original 1 fps keyframes accordingly. The full 20 fps datasets will be released in our future journal version.We strongly recommend using ProstaTDv2 for your experiments.
We plan to release the Bbox annotation and other resources at the end of April through the ICLR conference. Stay tuned!
[!WARNING]
In the ProstaTDv2 release (before Feb 6), we missed merging one category. Please re-download file using the same link as before (no need to re-apply).
<a id=“cross-validation”></a>
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。
数据缺失？
依托陕西、四川两大基地，我们与超过十家三甲医院建立直接合作关系，覆盖合规授权、采集治理、专业标注、数据交付的全流程，为AI医疗团队提供即拿即用的高质量临床数据。
⚡️ 需要数据支持或标注服务？立即联系我们获取专业方案。
