TOM 舌象分割工具 · TOM Tongue Segmentation

AI 驱动的中医舌象智能分割工具 · AI-powered tongue segmentation for TCM research

TOM app icon

GitHub Release Platform License

🇨🇳 中文 · 🇺🇸 English


🇨🇳 中文版

一款面向中医研究与临床的舌象分割桌面工具,自动从舌头照片中分离出舌体区域,支持单张精修和批量处理。

✨ 功能特性

  • 🎯 一键自动分割 —— 打开图片即自动识别舌头区域,通常 1–3 秒完成
  • 🎨 交互式微调 —— 通过点击 / 框选快速修正分割结果,每次标注后自动重算
  • 📁 批量处理 —— 整个文件夹一次性处理,可同时输出多种格式
  • 💾 三种输出格式 —— 透明背景 PNG · 二值掩膜 · 叠加可视化
  • 高精度模式 —— 可选 GrabCut 边界精修,更贴合舌体边缘
  • 🌏 中英双语界面 —— 一键切换中文 / English
  • 🪟 绿色安装 —— 双击安装即用,无需 Python / CUDA 环境

📥 下载安装

可直接从本仓库下载,或前往 GitHub Releases 页面:

文件 大小 平台
TOM-Tongue-Segmentation-v3.2.0.exe 565 MB Windows 10 / 11(64 位)
TOM-Tongue-Segmentation-v3.2.0-arm64.dmg 354 MB macOS 11+(Apple Silicon)

Windows:双击 .exe 运行安装向导,可选择:

  • 语言:中文 / English(界面语言可在安装后随时切换)
  • 安装位置:默认 %LOCALAPPDATA%\Programs\TOM舌象分割工具\
  • 桌面快捷方式:是否创建

安装完成后从开始菜单或桌面快捷方式启动。

macOS:双击 .dmg 挂载,将 TOM Tongue Segmentation.app 拖入 Applications 文件夹即可。首次启动时由于未经 Apple 公证,可能提示「无法验证开发者」——请在 Applications右键点击应用 → 选择「打开」→ 在弹窗中再次点「打开」即可正常运行,之后双击启动不再提示。

🚀 使用方法

单张图片处理

单张图片界面

  1. 点击 打开图片,选择一张舌头照片
  2. 等待 1–3 秒,应用自动完成分割
  3. 如对结果不满意,可进行交互式微调:
    • Shift + 左键拖动:框选舌头区域(最准确)
    • 左键点击:标记舌头(前景点)
    • 右键点击:标记背景
    • 每次标注后会自动重新分割
  4. 满意后点击 保存结果,选择输出文件夹
  5. 自动生成三种输出:透明背景 PNG · 二值掩膜 · 叠加可视化图

批量处理

批量处理界面

  1. 切换到 批量处理 标签页
  2. 选择 输入文件夹(内含待处理的舌头图片)
  3. 选择 输出文件夹
  4. 勾选需要的输出类型(抠图 / 掩膜 / 叠加图)
  5. 可选:勾选 高精度模式(GrabCut 边界精修,每张多耗时约零点几秒)
  6. 点击 开始批量分割
  7. 进度条和日志会实时显示处理状态

支持的图片格式:.jpg .jpeg .png .bmp .tif .tiff .webp

切换语言

点击右上角 EN / 中 按钮即可在中英文之间切换。

📋 系统要求

  • 操作系统:Windows 10 / 11(64 位)· macOS 11 Big Sur 或更高(Apple Silicon,M1/M2/M3/M4)
  • 内存:4 GB RAM 以上(推荐 8 GB)
  • 磁盘空间:约 1.5 GB
  • 加速:Windows 可选 GPU 加速 · macOS 自动启用 Apple MPS 加速(CPU 也能正常运行)

🇺🇸 English

A desktop tongue-segmentation tool for TCM research and clinical use. Automatically isolates the tongue region from a photograph, with support for single-image fine-tuning and batch processing.

✨ Features

  • 🎯 One-click auto segmentation — Tongue region is identified automatically on opening, typically in 1–3 seconds
  • 🎨 Interactive fine-tuning — Click / box-select to correct results; re-segments after every edit
  • 📁 Batch processing — Process an entire folder at once, output multiple formats simultaneously
  • 💾 Three output formats — Transparent PNG cutout · Binary mask · Overlay visualization
  • High-accuracy mode — Optional GrabCut boundary refinement for tighter edges
  • 🌏 Bilingual UI — Switch between English and 中文 with one click
  • 🪟 Standalone install — Double-click installer, no Python / CUDA setup required

📥 Download & Install

Download directly from this repository, or visit the GitHub Releases page:

File Size Platform
TOM-Tongue-Segmentation-v3.2.0.exe 565 MB Windows 10 / 11 (64-bit)
TOM-Tongue-Segmentation-v3.2.0-arm64.dmg 354 MB macOS 11+ (Apple Silicon)

Windows: Double-click the .exe to run the installer wizard, which lets you choose:

  • Language: 中文 / English (UI language can be switched anytime after install)
  • Install location: defaults to %LOCALAPPDATA%\Programs\TOM Tongue Segmentation\
  • Desktop shortcut: yes / no

Launch from the Start menu or desktop shortcut after install.

macOS: Double-click the .dmg to mount it, then drag TOM Tongue Segmentation.app into the Applications folder. Because the bundle is not Apple-notarized, the first launch may show "cannot be verified". To get past it, right-click the app in Applications → Open → Open in the confirmation dialog. After that first launch, double-clicking works normally.

🚀 How to Use

Single Image

Single image UI

  1. Click Open Image and pick a tongue photo
  2. Wait 1–3 seconds — the app segments automatically
  3. If you want to refine the result:
    • Shift + left-drag: box the tongue region (most accurate)
    • Left click: add a foreground (tongue) point
    • Right click: mark background
    • Re-segmentation runs after every edit
  4. When satisfied, click Save Results and pick an output folder
  5. Three outputs are produced: transparent PNG · binary mask · overlay visualization

Batch Folder

Batch UI

  1. Switch to the Batch Folder tab
  2. Pick an Input folder (containing tongue images)
  3. Pick an Output folder
  4. Tick which output types you want (cutout / mask / overlay)
  5. Optional: enable High accuracy (GrabCut edge refine, adds a fraction of a second per image)
  6. Click Start batch
  7. Progress bar and log update in real time

Supported image formats: .jpg .jpeg .png .bmp .tif .tiff .webp

Switching Language

Click the EN / 中 button at the top-right to toggle between English and Chinese.

📋 System Requirements

  • OS: Windows 10 / 11 (64-bit) · macOS 11 Big Sur or newer (Apple Silicon, M1/M2/M3/M4)
  • RAM: 4 GB or more (8 GB recommended)
  • Disk: ~1.5 GB
  • Acceleration: Optional GPU on Windows · Apple MPS used automatically on macOS (CPU also works)

📖 引用 · Citation

如本工具对您的研究有帮助,欢迎引用我们的文章。 If this tool helps your research, please cite our paper:

Xie, J., Zhang, Z., Poudel, B., Guo, C., Yu, Y., An, G., ... & Xu, D. (2026). TOM: An open-source tongue segmentation method with multi-teacher distillation and task-specific data augmentation. Expert Systems with Applications, 131499.

🔗 View on ScienceDirect

🏛️ 研发团队 · Developed by

Digital Biology Laboratory (DBL) University of Missouri

如本工具对您有帮助,欢迎给我们一个 ⭐ Star 以获取后续更新通知。 If this tool helps you, a ⭐ star is appreciated — and you'll get notified of future updates.

📄 版权 · Copyright

© 2026 Digital Biology Laboratory · University of Missouri

Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support