📸 Image Label IA

Image Label IA is a high-performance desktop application designed specifically for photographers. It automates image tagging and metadata generation using state-of-the-art machine learning models (ConvNeXt and DETR), run directly on your local device via Tauri, Angular, and Rust/Python.

---

📖 Table of Contents

• About the Project
• Key Features
• Tech Stack
• Prerequisites
• Installation & Setup
• Configuration
• Usage
• Developer Guides
• Contributing
• Security
• License

---

🎨 About the Project

Photographers often spend hours manually keywording, organizing, and tagging photos. Image Label IA solves this by performing locally run AI-assisted classification and object detection. By leveraging ONNX Runtime in Rust (ort), the application analyzes images instantly and updates metadata using standard EXIF/IPTC/XMP headers, without sending your private photographs to third-party cloud services.

Application Screenshots

Main Dashboard	Image Details & AI Tags

Batch Processing	Run Completion Report

---

✨ Key Features

• 🚀 Hybrid Desktop Architecture: Powered by Tauri v2 and Angular 20 for a lightweight, secure, and beautiful user interface.
• 🤖 Offline Local AI Inference: Runs ConvNeXt for image classification and DETR for object detection offline via ONNX Runtime (ort in Rust).
• ⚡ Advanced Gallery UX: Smooth progressive thumbnail rendering with lazy loading and a thread-concurrency decoder cache to prevent interface lag.
• 📁 Workspace Persistence: Remembers your last-viewed directory and restores the navigation path seamlessly on relaunch.
• 🗂️ Collapsible Panels: Toggleable left and right sidebars (Folder Tree and Metadata details) to maximize photo viewing estate.
• 🏷 Metadata Synchronization: Automatically writes confirmed AI keyword labels back to image standard IPTC headers.
• ⚙️ Configurable AI Confidence: Adjust confidence thresholds in the batch config modal to fine-tune auto-tag matching.

---

🛠 Tech Stack

• Frontend: Angular 20, TypeScript, Vanilla CSS
• App Wrapper & Core: Tauri v2 (Rust)
• AI Inference Backend: Rust ort (ONNX Runtime wrapper)
• Model Processing Scripts: Python 3 (using torch, transformers, huggingface-hub)
• Metadata Management: rexiv2 (GExiv2 wrapper)

---

⚙️ Prerequisites

Before you begin, ensure you have the following installed on your developer machine:

1. Rust Toolchain: Install rustup to manage Rust compiler versions.

   curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh

2. Node.js: Long-Term Support (LTS) version (v18 or higher recommended).
3. pnpm: Fast, disk-efficient package manager.

   npm install -g pnpm

4. Python 3.9+: For running model conversion/download scripts.
5. System Dependencies (Linux only): Tauri and rexiv2 (exiv2) require system packages. On Debian/Ubuntu:

   sudo apt-get update
   sudo apt-get install -y libglib2.0-dev libgexiv2-dev libwebkit2gtk-4.1-dev build-essential curl wget file libxdo-dev libssl-dev libayatana-appindicator3-dev librsvg2-dev

---

🚀 Installation & Setup

Follow these steps to clone the repository and set up your local development environment:

1. Clone the Repository

git clone https://github.com/lakescorp/ImageLabelIA.git
cd ImageLabelIA

2. Install Node Dependencies

Use pnpm to install frontend packages:

pnpm install

3. Setup Python Virtual Environment & Install Scripts

Create a virtual environment and install dependencies to fetch/convert the ONNX models:

python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt  # Or manually: pip install huggingface-hub torch transformers

4. Fetch the ONNX Model

Run the download script to retrieve the pre-converted ConvNeXt model and label set from HuggingFace Hub:

python download_model.py

This saves:

• src-tauri/resources/convnext.onnx
• src-tauri/resources/convnext_labels.json

5. Launch the Application in Dev Mode

Start the Angular dev server and launch the Tauri window:

pnpm tauri dev

---

⚙️ Configuration

The application reads configurations from environment variables.

1. Copy the environment template:

   cp .env.example .env

2. Customize the settings inside .env to suit your system requirements (e.g. changing inference execution providers from cpu to cuda). Refer to .env.example for detailed comments.

---

📸 Usage

Once the application is running:

1. Select Folder: Click "Process Folder" to open a native directory picker and select your photo directory.
2. Configure Settings: Select the desired confidence thresholds and tag processing options in the sidebar.
3. Run AI Engine: Click "Start Tagging". The UI will display a real-time progress bar.
4. Accept / Reject Tags: Click thumbnails to inspect labels. Confirm or modify tags before applying them to IPTC metadata.
5. Save to Disk: Apply suggestions to write metadata tags directly into image files.

---

🛠 Developer Guides

Project Structure

├── .github/                 # GitHub CI/CD workflows, Pull Request & Issue Templates
├── .vscode/                 # IDE Configuration
├── src/                     # Angular Single Page App (Frontend)
│   ├── app/                 # Components and routing
│   └── assets/              # App static files
├── src-tauri/               # Rust Tauri Core (Backend)
│   ├── resources/           # Pre-compiled ONNX models and category tags
│   ├── src/                 # Main Tauri entry point, file system, and inference logic
│   └── Cargo.toml           # Rust Cargo dependencies
├── download_model.py        # Helper script to pull models from HF Hub
├── package.json             # NPM package definitions
└── README.md                # This file

---

🤝 Contributing

Contributions make the open-source community an amazing place to learn, inspire, and create. Please check out CONTRIBUTING.md to understand our branching strategy, linting guidelines, and code submission protocols.

---

🔒 Security

For reporting security vulnerabilities or issues regarding data leaks, please consult our SECURITY.md. Please do not open public GitHub issues for security reports.

---

📜 License

Distributed under the MIT License. See LICENSE for more details.