Skip to content

dvtlab/TagiFLY

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

10 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

TagiFLY v2.0.0 🚀

AI Labelling Tool for Computer Vision

License: MIT Electron Version

Open source, lightweight, and focused only on what you need.

TagiFLY v2.0.0 is a cross-platform desktop application for AI computer vision data labeling.

📸 Screenshots

Main Screen Annotation Example
Main Annotation
Export Window Dark Mode Light Mode
Export Dark Light

✨ Features

🎯 Annotation Tools

  • Bounding Box: Rectangle annotations for object detection
  • Polygon: Complex shape annotations with precise boundaries
  • Point: Single point annotations for key features
  • Keypoint/Pose: 17-point human pose estimation annotations
  • Mask Paint: Brush-based segmentation annotations
  • Polyline: Line and path annotations

📤 Multiple Export Formats

  • JSON: General-purpose format with full metadata
  • YOLO: Machine learning format for object detection
  • COCO: Microsoft COCO dataset format
  • Pascal VOC: XML format for computer vision

🔧 Advanced Features

  • Label Management: Export/import label configurations
  • Performance Optimized: Cached rendering and smooth interactions
  • Cross-Platform: Windows, macOS, and Linux support
  • Professional Canvas: High-performance image rendering

⚠️ Note

This is the **second release **.
It may contain bugs or missing features. Please share your feedback so we can improve it together. 🚀

🚀 Quick Start

Prerequisites

  • Node.js 16+
  • npm or yarn

Installation

  1. Clone the repository
    git clone https://github.com/dvtlab/tagifly.git
    cd tagifly
  2. Install dependencies
    npm install
  3. Start the application
    npm start

📖 User Guide

Getting Started

  1. Select Folder: Click "Select Folder" to load your images
  2. Choose Annotation Tool: Select from 6 professional annotation tools
  3. Add Labels: Create custom labels or use predefined ones
  4. Annotate: Click and drag to create annotations
  5. Export: Choose from 4 export formats (JSON, YOLO, COCO, Pascal VOC)

Keyboard Shortcuts

Shortcut Action
B Bounding Box tool
P Polygon tool
O Point tool
K Keypoint/Pose tool
M Mask Paint tool
L Polyline tool
F Fit to screen
1-9 Quick label assignment
Ctrl+Z Undo
Ctrl+Y Redo
Delete Remove selected annotation
ESC Cancel current operation

🆕 What's New in v2.0.0

🎯 Revolutionary Improvements

🎨 User Experience Revolution

  • ✨ Smart Notifications: Moved to center header - no more UI blocking
  • ✨ Dark Mode Perfection: Fixed invisible text in input fields and zoom indicators
  • ✨ Keyboard Harmony: Resolved all conflicts between shortcuts and input fields
  • ✨ Zoom Mastery: Fixed centering issues - zoom now works like Photoshop
  • ✨ Pose Guidance: Step-by-step instructions for 17-point human pose annotation

🔄 Annotation Tools Enhancement

  • 🎯 Keypoint/Pose Selection: Fixed drag and move - now works flawlessly
  • 🎯 Mask Paint Selection: Added proper selection handling for brush strokes
  • 🎯 ESC Key Intelligence: Enhanced pose drawing cancellation with cleanup
  • 🎯 Universal Undo/Redo: Fixed for all annotation types - never lose work again

📤 Brand New Label Management

  • 🆕 Label Export/Import: Save your label configurations as JSON files
  • 🆕 Professional Tab System: Two-tab interface (Labels & Import/Export)
  • 🆕 Smart Statistics: Real-time label count and last updated tracking
  • 🆕 Duplicate Prevention: Intelligent merging during label import
  • 🆕 Workflow Continuity: Continue projects across sessions

⚡ Performance Breakthrough

  • 🚀 Memory Optimization: Advanced caching for smooth performance
  • 🚀 Animation Smoothness: Enhanced UI transitions and interactions
  • 🚀 Error Resilience: Robust error handling throughout the application

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

👨‍💻 Author

TagiFLY v2.0.0 - Labelling Tool for AI Computer Vision 🚀

⭐ If TagiFLY helps you, don’t forget to leave a star!

Made with 💙 by dvtlab

📧 Contact: [email protected]