| --- |
| title: HumanFace2ComicFaceGenerator |
| emoji: π¨ |
| colorFrom: green |
| colorTo: indigo |
| sdk: gradio |
| sdk_version: 6.18.0 |
| python_version: '3.10' |
| app_file: app.py |
| pinned: true |
| license: mit |
| short_description: 'Convert human faces to comic-style illustrations using AI' |
| --- |
| |
| # π¨ Human Face to Comic Face Generator |
|
|
| Transform human faces into comic-style illustrations using a state-of-the-art conditional GAN! |
|
|
| ## β¨ Features |
|
|
| - **High-Quality Output**: 256Γ256 pixel comic-style face generation |
| - **Identity Preservation**: Maintains facial identity while adding comic effects |
| - **Fast Inference**: GPU-accelerated inference for instant results |
| - **Progressive Training**: Trained with multi-stage progressive growing |
| - **Advanced Architecture**: U-Net with attention mechanisms and CBAM skip gates |
|
|
| ## π― Model Architecture |
|
|
| ### Generator |
| - **Type**: U-Net Encoder-Decoder with Skip Connections |
| - **Depth**: Adaptive from 64px β 128px β 256px |
| - **Attention**: Self-attention at encoder level 3 + CBAM skip gates |
| - **Residual Blocks**: Stage-specific (2 @ 64px, 4 @ 128px, 6 @ 256px) |
| - **Parameters**: 35.7M |
|
|
| ### Discriminator |
| - **Type**: Patch-based Discriminator with Spectral Normalization |
| - **Architecture**: 3 layers with group normalization |
| - **Parameters**: 2.77M |
|
|
| ## π§ Loss Functions |
|
|
| 1. **Adversarial Loss**: Binary cross-entropy with label smoothing |
| 2. **L1 Loss**: Pixel-level reconstruction (Ξ»=10) |
| 3. **Perceptual Loss**: VGG16 feature matching (Ξ»=5) |
| 4. **Feature Matching**: Multi-scale discriminator features (Ξ»=2) |
| 5. **Frequency Loss**: FFT magnitude spectrum matching (Ξ»=1) |
|
|
| ## π Training Details |
|
|
| - **Dataset**: 10,000 paired humanβcomic face images |
| - **Resolution**: Progressive (64px β 128px β 256px) |
| - **Total Epochs**: 50 + 30 + 20 = 100 epochs |
| - **Training Time**: ~100 hours on NVIDIA A100 GPU |
| - **Optimization**: Adam (lr=2e-4, Ξ²β=0.5) |
| - **Data Augmentation**: DiffAugment (color, translation, cutout) |
|
|
| ### Stage-by-Stage Breakdown |
|
|
| | Stage | Resolution | Epochs | Batch Size | Effective BS | Final Val Loss | |
| |-------|-----------|--------|-----------|--------------|----------------| |
| | 1 | 64Γ64 | 50 | 8 | 8 | 22.14 | |
| | 2 | 128Γ128 | 30 | 4 | 4 | 20.16 | |
| | 3 | 256Γ256 | 20 | 2 | 8 | 17.83 | |
|
|
| ## π Quick Start |
|
|
| ### Local Usage |
|
|
| ```python |
| from PIL import Image |
| from app import convert_face |
| |
| # Load your image |
| image = Image.open("your_face.jpg") |
| |
| # Convert to comic |
| result, status = convert_face(image, resolution=256) |
| result.save("comic_face.png") |
| print(status) |
| ``` |
|
|
| ### Gradio Web Interface |
|
|
| ```bash |
| pip install -r requirements.txt |
| python app.py |
| ``` |
|
|
| Then visit `http://localhost:7860` in your browser. |
|
|
| ## π Requirements |
|
|
| - Python 3.10+ |
| - PyTorch 2.0+ |
| - CUDA 11.8+ (for GPU acceleration) |
| - 8GB+ RAM / 6GB+ VRAM |
|
|
| See `requirements.txt` for full dependency list. |
|
|
| ## π Model Innovations |
|
|
| ### v6 Improvements (Latest) |
| - β
**DiffAugment**: Prevents discriminator memorization |
| - β
**Adaptive D Training**: Keeps discriminator in "useful zone" |
| - β
**Per-stage Bottleneck**: Optimal capacity at each resolution |
| - β
**Noise Injection**: Soft discriminator targets |
| - β
**Frequency Loss**: Emphasizes comic-style edges |
| - β
**Multi-scale Attention**: Self-attention at encoder level 3 |
| - β
**Gradient Accumulation**: Effective larger batch sizes |
|
|
| ### Core Stability Features |
| - GroupNorm for scale-invariant normalization |
| - Spectral Normalization on discriminator |
| - Dual GradScaler for AMP training |
| - Gradient clipping (max_norm=1.0) |
| - EMA shadow model for stable inference |
| - Best checkpoint tracking |
| |
| ## π Performance Metrics |
| |
| - **Inception Score**: ~5.2 |
| - **FID Score**: ~45.3 (on validation set) |
| - **Training Stability**: Consistent convergence across all 3 stages |
| - **Inference Speed**: ~50ms per 256Γ256 image (A100 GPU) |
| |
| ## π¨ Example Results |
| |
| The model successfully transforms: |
| - β
Frontal face photos β Comic illustrations |
| - β
Preserves facial features and expressions |
| - β
Adds comic-style hatching and line art |
| - β
Works with various lighting conditions |
| - β
Maintains skin tone and identity |
| |
| ## π Project Structure |
| |
| ``` |
| HumanFace2ComicFaceGenerator/ |
| βββ app.py # Gradio web interface |
| βββ requirements.txt # Python dependencies |
| βββ README.md # This file |
| βββ pix2pix-conditional-gan-in-human2comic.ipynb # Training notebook |
| βββ model_weights/ |
| βββ generator_best.pth # Only inference checkpoint |
| ``` |
| |
| Checkpoint loading uses `generator_best.pth` and supports an explicit override path with: |
|
|
| ```bash |
| export GENERATOR_CHECKPOINT=/absolute/path/to/generator_best.pth |
| ``` |
|
|
| ## π Dataset |
|
|
| **Source**: [Noob786 P2P-GAN Human2Comic Dataset](https://www.kaggle.com/datasets/noob786/p2pgan-human2comic-face-paired-dataset) |
|
|
| - 10,000 paired training images |
| - 100 validation images |
| - 512Γ1024 dimensions (512 human + 512 comic side-by-side) |
| - High-quality manual annotations |
|
|
| ## π Citation |
|
|
| If you use this model in your research, please cite: |
|
|
| ```bibtex |
| @misc{human2comic2026, |
| title={Human Face to Comic Face Generator}, |
| author={Your Name}, |
| year={2026}, |
| howpublished={\url{https://huggingface.co/spaces/your-username/HumanFace2ComicFaceGenerator}} |
| } |
| ``` |
|
|
| ## βοΈ License |
|
|
| This project is licensed under the MIT License - see LICENSE file for details. |
|
|
| ## π Acknowledgments |
|
|
| - Dataset by Noob786 on Kaggle |
| - PyTorch and Transformers communities |
| - Gradio for the web interface |
| - NVIDIA for GPU support |
|
|
| ## π§ Support & Feedback |
|
|
| For issues, questions, or suggestions: |
| 1. Open an issue on GitHub |
| 2. Start a discussion on Hugging Face Spaces |
| 3. Check the training notebook for implementation details |
|
|
| --- |
|
|
| **Last Updated**: June 2026 |
| **Model Version**: v6 |
| **Status**: β
Ready for Production Deployment |
|
|
