README.md

metadata

title: Labelizer - AI Image Labeling Tool
emoji: 🖼️
colorFrom: blue
colorTo: purple
sdk: gradio
sdk_version: 6.0.2
app_file: app.py
pinned: false
license: mit
python_version: '3.12'

🖼️ Labelizer - AI Image Labeling Tool

An intelligent image labeling tool that uses Florence-2 vision-language model to automatically generate detailed descriptions for your images. Perfect for creating labeled datasets for machine learning projects.

✨ Features

• 🤖 AI-Powered Labeling: Uses advanced Florence-2 model for accurate image descriptions
• 📊 Batch Processing: Label multiple images at once with progress tracking
• ✏️ Manual Editing: Edit generated labels to fit your specific needs
• 📦 Flexible Export: Download datasets with organized folder structure or flat format
• 🎨 User-Friendly Interface: Clean, intuitive Gradio interface with emoji-enhanced navigation

🚀 How to Use

1. Upload Images: Click "📁 Upload images" to select multiple image files
2. Generate Labels:
   • Click "✨ Generate label" below individual images
   • Or click "🏷️ Labelize all images" for batch processing
3. Review & Edit: Modify any generated labels as needed
4. Download: Create and download your labeled dataset as a ZIP file

🛠️ Technical Details

• Model: Florence-2-large-hf for vision-language understanding
• Framework: Gradio with ZeroGPU support
• Supported Formats: JPG, PNG, GIF, BMP, TIFF, WebP
• Export Options: Organized folders (images/ + labels/) or flat structure

📋 Supported Tasks

The tool supports various captioning tasks:

• <MORE_DETAILED_CAPTION>: Comprehensive image descriptions
• <DETAILED_CAPTION>: Detailed but concise descriptions
• <CAPTION>: Basic image captions

🎯 Use Cases

• Machine Learning: Create labeled datasets for computer vision tasks
• Content Management: Organize image collections with descriptions
• Accessibility: Generate alt-text for images
• Research: Prepare datasets for academic projects

⚡ Performance

• Optimized for GPU acceleration with ZeroGPU
• Efficient batch processing for large datasets
• Lazy loading to minimize resource usage

---

Built with ❤️ using Gradio and Florence-2