Image Class Selector

A desktop application built with Tauri, React, and TypeScript for creating YOLO annotations. Draw rectangles on images and assign them to custom classes.

Features

• Load Images: Select folders of images for annotation
• Load Classes: Define custom object classes
• Draw Rectangles: Click and drag to create selection rectangles
• Assign Classes: Select rectangles and assign them to classes
• Export YOLO: Export annotations in YOLO format for machine learning
• Clear Annotations: Remove rectangles from current image or all images
• System Menu: File menu with Select Folder, Load Classes, and Quit
• Keyboard Shortcuts: Cmd+O (Select Folder), Cmd+L (Load Classes)

How to Use

1. Load Classes: Click "Load Classes" to select a text file with class names (one per line)
2. Select Folder: Click "Select Folder" to choose a directory with images
3. Annotate: Click thumbnails to view images, then click and drag to create rectangles
4. Assign Classes: Select rectangles and choose a class from the dropdown
5. Export: Click "Export YOLO" to save annotations in YOLO format
6. Clear: Use "Clear This" or "Clear All Images" to remove annotations

Keyboard Shortcuts

• Cmd+O (macOS) / Ctrl+O (Windows/Linux): Select Folder
• Cmd+L (macOS) / Ctrl+L (Windows/Linux): Load Classes

Technical Stack

• Frontend: React + TypeScript + Vite
• UI Framework: Ant Design
• Canvas Library: React Konva (Konva.js)
• Desktop Framework: Tauri v2
• Backend: Rust

Installation

Prerequisites

• Node.js (v18 or higher)
• Rust (latest stable)

Development

# Install dependencies
npm install

# Start development server
npm run tauri dev

# Build for production
npm run tauri build

Windows Build

Use the provided build scripts:

build-windows.bat

or

.\build-windows.ps1

Distribution

Pre-built apps are available in the distributables/ folder:

• macOS: Image Class Selector_0.1.0_x64.dmg
• Windows: Use build scripts to create distributables