Image Tagging

A Python application for tagging and categorizing images.

Description

This project provides functionality for automatically or manually tagging images with relevant labels and categories.

Installation

1. Clone the repository:

git clone <repository-url>
cd ImageTagging

2. Create a virtual environment (recommended):

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies:

pip install -r requirements.txt

Usage

Running EDA (Exploratory Data Analysis)

Before building any models, it's recommended to run the EDA pipeline to understand your data:

python run_eda.py

This will:

1. Load all JSON files from entities_dataset_v2/
2. Analyze schema and data structure
3. Analyze tag frequencies and distributions
4. Analyze image-level statistics
5. Generate visualizations
6. Save summary reports to eda_output/

Outputs:

• Console reports with key insights
• CSV files with detailed statistics (tag frequencies, co-occurrences, etc.)
• JSON files with summary statistics
• Visualization plots (PNG files)

Running the Main Application

Run the main application:

python main.py

Project Structure

ImageTagging/
├── main.py                      # Main entry point
├── run_eda.py                   # EDA pipeline script
├── requirements.txt             # Python dependencies
├── entities_dataset_v2/         # JSON dataset files
├── eda/                         # EDA modules
│   ├── __init__.py
│   ├── data_loader.py          # Data loading and normalization
│   ├── schema_analyzer.py       # Schema analysis
│   ├── tag_analyzer.py          # Tag/entity analysis
│   ├── image_analyzer.py        # Image-level analysis
│   └── visualizer.py            # Visualization generation
├── eda_output/                  # Generated EDA outputs (created after running)
└── README.md                    # This file

Development

Add your image tagging logic in main.py or create additional modules as needed.

License

[Add your license here]

Contributing

[Add contribution guidelines here]