NeoNude

GAN-based Image Transformation
A pix2pixHD architecture using a divide-et-impera approach for image transformation.

---

✨ Features

• Auto Resize — Automatically resizes any input to 512x512 and restores original dimensions on output
• 3-Phase GAN Pipeline — Clothing mask → Anatomical detection → Final generation
• OpenCV Transforms — Color correction, mask refinement, and mask finalization
• pix2pixHD Architecture — Global generator with residual blocks and instance normalization
• Simple CLI — Easy input/output with minimal arguments

---

📋 Requirements & Installation

Requirements

Component	Requirement
Python	3.7 or higher
PyTorch	1.7 or higher
OpenCV	4.5+ (headless)
Pillow	8.0+
NumPy	1.19+

Installation

# Clone the repository
git clone https://github.com/fahimahamed1/NeoNude.git
cd NeoNude

# Install dependencies
pip install -r requirements.txt

Model Checkpoints

Place the model weight files in the checkpoints/ directory:

checkpoints/
├── cm.lib    (~700 MB)
├── mm.lib    (~700 MB)
└── mn.lib    (~700 MB)

📖 How to Use

CLI

# Default: reads input.png, writes output.png
python main.py

# Custom input/output paths
python main.py -i photo.jpg -o result.png

# Help
python main.py --help

⚙️ Pipeline Architecture

Instead of a single network, the problem is split into 3 sub-problems:

1. Mask generation — Identifies clothing regions
2. Anatomical attribute detection — Produces an abstract body map
3. Final image generation — Creates the output from the refined mask

Pipeline Phases

Phase	Type	Description
0	OpenCV	Color correction and normalization
1	GAN	Clothing mask generation (cm.lib)
2	OpenCV	Mask refinement
3	GAN	Anatomical detail detection (mm.lib)
4	OpenCV	Mask finalization with body annotations
5	GAN	Final image generation (mn.lib)

📂 Project Structure

NeoNude/
├── main.py                      # CLI entry point
├── requirements.txt             # Python dependencies
├── README.md
├── src/                         # Core package
│   ├── __init__.py              # Package metadata
│   ├── config.py                # Pipeline configuration (Options)
│   ├── pipeline.py              # Pipeline orchestrator (process function)
│   ├── model.py                 # GAN model, dataset, utilities
│   └── transforms/              # OpenCV image transforms
│       ├── __init__.py
│       ├── annotation.py        # Body part data class
│       ├── correct.py           # Phase 0: color correction
│       ├── maskref.py           # Phase 2: mask refinement
│       └── maskfin.py           # Phase 4: mask finalization
└── checkpoints/                 # Model weight files (not tracked)

---

🛡️ Security Notes

⚠️ This tool is intended for:

• Personal use and educational purposes
• Research in GAN-based image transformation
• Academic study of pix2pixHD architecture

Do NOT use for:

• Generating non-consensual intimate imagery
• Any form of harassment or exploitation
• Any illegal activities

Always ensure proper authorization and ethical use before processing any images.

---

📄 License

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

---

👨‍💻 Author

Fahim Ahamed

---

⭐ Support

If you find this project useful, please consider giving it a star! 🌟

---

Made with ❤️ for the open-source community