登录

cucim

rapidsaiApache-2.0活跃维护

基于 CUDA 的 GPU 加速图像处理库，专为多维图像和数字病理、显微成像等大规模图像分析场景提速。

NVIDIA cuCIM - GPU-accelerated image processing library

v26.10.00发布于 2026-10-07 19:33 UTC474star93fork同步于 2026-11-10 17:59 UTC
GPU加速图像处理数字病理医学影像多维图像
对比 GitHub 官网computer-visioncudadigital-pathologygpuimage-analysis

怎么上手

由 README 提炼，执行前请确认
原文 
conda create -n cucim -c rapidsai -c conda-forge cucim cuda-version=<CUDA version>
pip install cucim-cu12
硬件要求按 README 说明

README 未说明具体显存要求；安装包按 CUDA 版本区分（CUDA 12 或 13），需 NVIDIA GPU 与对应 CUDA 环境。

需要 CUDA
下载：源码 zip·GitHub Releases（安装包）本站不镜像安装包

项目说明

由 README 自动翻译提炼
没人工点评？写下你的经验

NVIDIA 的 GPU 加速多维图像处理库，专为病理切片和遥感大图做 I/O 与运算提速

基于 CUDA 的 GPU 加速图像处理库，专为多维图像和数字病理、显微成像等大规模图像分析场景提速。

核心能力

  • 用 GPU 加速图像处理与计算机视觉基础算子
  • 支持读取 SVS、Philips TIFF 及多种压缩的多分辨率 TIFF
  • 提供与 OpenSlide 对齐的 Pythonic API
  • cuslide2 插件用 nvImageCodec 做 GPU TIFF 解码并支持瓦片缓存
  • 支持异步批量解码与更广的格式覆盖

容易踩的坑

conda 安装需把 换成 12.0+ 的具体版本号；PyPI 包按 CUDA 版本分开，装错 cucim-cu12/cu13 会不匹配；cuslide2 需手动设置 ENABLE_CUSLIDE2=1 才启用，且旧 cuslide 插件将在 26.04 起弃用；跑 notebook 需先用 Docker 下载示例图像。

内容由模型根据 README 生成，可能不完整或不准确，以官方文档为准。看原文 README

最近发版

5 个版本v26.10.00
全部 release
  1. v26.10.00最新正式版2026-10-07 · 2026-10-07 19:33 UTC

    What's Changed

    🐛 Bug Fixes

    📖 Documentation

    🛠️ Improvements

    Full Changelog: https://github.com/rapidsai/cucim/compare/v26.10.00a...v26.10.00

  2. v26.08.002026-08-05 · 2026-08-05 17:53 UTC
    查看更新说明

    What's Changed

    🚨 Breaking Changes

    🐛 Bug Fixes

    📖 Documentation

    🛠️ Improvements

    Full Changelog: https://github.com/rapidsai/cucim/compare/v26.08.00a...v26.08.00

  3. v26.06.002026-06-03 · 2026-06-03 18:16 UTC
    查看更新说明

    What's Changed

    🚨 Breaking Changes

    🛠️ Improvements

    New Contributors

    Full Changelog: https://github.com/rapidsai/cucim/compare/v26.06.00a...v26.06.00

  4. v26.04.002026-04-08 · 2026-04-08 15:26 UTC
    查看更新说明

    What's Changed

    🚨 Breaking Changes

    🐛 Bug Fixes

    📖 Documentation

    🛠️ Improvements

    Full Changelog: https://github.com/rapidsai/cucim/compare/v26.04.00a...v26.04.00

  5. v26.02.012026-03-04 · 2026-03-04 01:41 UTC
    查看更新说明

    What's Changed

    🛠️ Improvements

    Full Changelog: https://github.com/rapidsai/cucim/compare/v26.02.00...v26.02.01

用户点评

登录用户的真实使用记录，可按版本和硬件筛选
共 0 条

    还没有人点评。第一个说出真实体验的人最有价值。

    README 原文

    未翻译
    在 GitHub 打开

    NVIDIA cuCIM

    NVIDIA cuCIM (pronounced "koo-sim", see here) is an open-source, accelerated computer vision and image processing software library for multidimensional images used in biomedical, geospatial, material and life science, and remote sensing use cases.

    cuCIM offers:

    • Enhanced Image Processing Capabilities for large and n-dimensional tag image file format (TIFF) files
    • Accelerated performance through Graphics Processing Unit (GPU)-based image processing and computer vision primitives
    • A Straightforward Pythonic Interface with Matching Application Programming Interface (API) for Openslide

    cuCIM supports the following formats:

    • Aperio ScanScope Virtual Slide (SVS)
    • Philips TIFF
    • Generic Tiled, Multi-resolution RGB TIFF files with the following compression schemes:
      • No Compression
      • JPEG
      • JPEG2000
      • Lempel-Ziv-Welch (LZW)
      • Deflate

    NOTE: For the latest stable README.md ensure you are on the main branch.

    Developer Page

    Blogs

    Webinars

    Documentation

    Release notes are available on our wiki page.

    Install cuCIM

    Conda

    Conda (stable)
    conda create -n cucim -c rapidsai -c conda-forge cucim cuda-version=`<CUDA version>`

    <CUDA version> should be 12.0+ (e.g., 12.0, etc.)

    Conda (nightlies)
    conda create -n cucim -c rapidsai-nightly -c conda-forge cucim cuda-version=`<CUDA version>`

    <CUDA version> should be 12.0+ (e.g., 12.0, etc.)

    PyPI

    Install for CUDA 12:

    pip install cucim-cu12

    Install for CUDA 13:

    pip install cucim-cu13

    Notebooks

    Please check out our Welcome notebook (NBViewer)

    Downloading sample images

    To download images used in the notebooks, please execute the following commands from the repository root folder to copy sample input images into notebooks/input folder:

    (You will need Docker installed in your system)

    ./run download_testdata

    or

    mkdir -p notebooks/input
tmp_id=$(docker create gigony/svs-testdata:little-big)
docker cp $tmp_id:/input notebooks
docker rm -v ${tmp_id}

    cuslide2 Plugin

    Deprecation Notice: Starting with 26.04, the original cuslide plugin will be deprecated in favor of cuslide2 with removal of cuslide planned for 26.08. cuslide2 uses nvImageCodec for GPU-accelerated TIFF decoding and adds tile-level caching, async batch decoding, and broader format support. Please plan to migrate existing workflows to cuslide2.

    Enabling cuslide2

    To enable the cuslide2 plugin, set the ENABLE_CUSLIDE2 environment variable:

    ENABLE_CUSLIDE2=1 python your_script.py

    Test Scripts

    cuCIM provides test scripts to verify cuslide2 functionality with different TIFF formats:

    Aperio SVS files:

    # Download a sample SVS file and run the test
python scripts/test_aperio_svs.py --download

# Or test with your own SVS file
ENABLE_CUSLIDE2=1 python scripts/test_aperio_svs.py /path/to/your/file.svs

    Philips TIFF files:

    # Test with a Philips TIFF file
ENABLE_CUSLIDE2=1 python scripts/test_philips_tiff.py /path/to/your/philips.tiff

# List available test options
python scripts/test_philips_tiff.py --help

    Build/Install from Source

    See build instructions.

    Contributing Guide

    Contributions to cuCIM are more than welcome! Please review the CONTRIBUTING.md file for information on how to contribute code and issues to the project.

    Acknowledgments

    Without awesome third-party open source software, this project wouldn't exist.

    Please find LICENSE-3rdparty.md to see which third-party open source software is used in this project.

    License

    Apache-2.0 License (see LICENSE file).

    Copyright (c) 2020-2026, NVIDIA CORPORATION.

    相关项目

    同分类里 star 最高的项目
    对比全部 →
    opencv
    opencv·C++

    开源计算机视觉库，提供图像处理、视频分析和深度学习推理的 C++ 接口，是视觉应用开发的基础设施。

    计算机视觉+4图像处理
    91525+4957k5.1.02026-11-10 12:05 UTC
    upscayl
    upscayl·TypeScript

    跨平台的开源 AI 图片放大工具，基于 ESRGAN 等模型把低分辨率图片无损放大并增强细节，有图形界面，普通用户也能直接上手。

    图像超分+4图像处理
    52174+2942.7kv2.15.02026-10-05 19:13 UTC
    supervision
    roboflow·Python

    一套可复用的计算机视觉工具库，提供目标检测、实例分割、分类等任务的标注处理、指标计算与可视化组件，帮开发者省去重复造轮子的时间。

    计算机视觉+4图像处理
    51582+494.9k0.31.52026-11-10 15:41 UTC
    sharp
    lovell·JavaScript

    基于 libvips 的高性能 Node.js 图像处理库，能快速完成 JPEG、PNG、WebP、AVIF、TIFF 的缩放裁剪与格式转换，适合服务端批量处理图片。

    图像处理+4图像处理
    32839+101.4kv0.35.52026-11-10 15:57 UTC