|
|
--- |
|
|
license: mit |
|
|
tags: |
|
|
- image-generation |
|
|
- unet |
|
|
- polygon-coloring |
|
|
- computer-vision |
|
|
--- |
|
|
|
|
|
# 🎨 Color Polygon UNet Model |
|
|
|
|
|
This model is a conditional U-Net trained to generate filled polygon images based on both **shape** and **color**. It takes two inputs: |
|
|
- A binary polygon mask image (e.g., a triangle, star, etc.) |
|
|
- A color condition (e.g., red, blue, yellow) |
|
|
|
|
|
It outputs a colorized version of the shape according to the given color prompt. |
|
|
|
|
|
## 🧠 Model Architecture |
|
|
|
|
|
- **Backbone**: U-Net with encoder-decoder structure |
|
|
- **Input channels**: 6 (3 for shape mask, 3 for color hint) |
|
|
- **Output**: RGB image of the filled shape |
|
|
|
|
|
## 📁 Files |
|
|
|
|
|
- `color_polygon_unet.pth`: Trained PyTorch model weights |
|
|
|
|
|
## 🔧 How to Use |
|
|
|
|
|
```python |
|
|
from huggingface_hub import hf_hub_download |
|
|
import torch |
|
|
|
|
|
# Download the model file |
|
|
model_path = hf_hub_download( |
|
|
repo_id="your-username/color-polygon-unet-model", |
|
|
filename="color_polygon_unet.pth" |
|
|
) |
|
|
|
|
|
# Load the model (assuming you have the UNet class defined) |
|
|
model = UNet(in_channels=6, out_channels=3) |
|
|
model.load_state_dict(torch.load(model_path, map_location='cpu')) |
|
|
model.eval() |
|
|
``` |
|
|
|
|
|
## 🧪 Intended Use |
|
|
This model is designed for: |
|
|
- Educational purposes |
|
|
- Learning conditional image generation |
|
|
- Demonstrating shape and color-controlled generative models |
|
|
|
|
|
## 🛑 Limitations |
|
|
- Trained on synthetic polygon data |
|
|
- Not suitable for real-world generalization |
|
|
- Only supports limited shapes and colors as trained (e.g., red star, blue circle) |
|
|
|
|
|
## 📜 License |
|
|
MIT License |
|
|
|
|
|
## 👤 Author |
|
|
Vighnesh M S |
|
|
|
|
|
|
