Improve output layout

README.md

@@ -9,30 +9,104 @@ app_file: app.py
 pinned: false
 ---
 
+<div align="center">
+
 # ClothQuill - AI Clothing Inpainting
 
-This Space allows you to inpaint clothing on images using Stable Diffusion. Upload an image, provide a prompt describing the clothing you want to generate, and get multiple inpainted results.
+[![Hugging Face Spaces](https://img.shields.io/badge/%F0%9F%A4%97%20Hugging%20Face-Spaces-blue)](https://huggingface.co/spaces/bismay/ClothQuill)
+[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
+[![Python 3.8+](https://img.shields.io/badge/python-3.8+-blue.svg)](https://www.python.org/downloads/release/python-380/)
+
+🎨 An AI-powered tool for seamlessly editing and modifying clothing in images using state-of-the-art deep learning models.
+
+[Demo](#demo) • [Features](#features) • [Installation](#installation) • [Usage](#usage) • [Examples](#examples) • [Technical Details](#technical-details)
+
+</div>
+
+## Demo
+
+ClothQuill is available as a Hugging Face Space! Try it out here: [ClothQuill Demo](https://huggingface.co/spaces/bismay/ClothQuill)
+
+## Features
+
+- 🎯 **Precise Clothing Detection**: Automatically identifies and segments different clothing items
+- 🎨 **Interactive Selection**: Choose specific clothing parts to modify
+- 🔄 **Multiple Variations**: Generate multiple inpainting results for each prompt
+- 🎛️ **Adjustable Controls**: Fine-tune mask dilation for better blending
+- 🖼️ **High-Quality Output**: Maintains image quality with advanced upscaling
+- 🚀 **Real-time Preview**: See segmentation masks and selected regions in real-time
+
+## Installation
+
+```bash
+# Clone the repository
+git clone https://huggingface.co/spaces/bismay/ClothQuill
+cd ClothQuill
+
+# Install dependencies
+pip install -r requirements.txt
 
-## How to Use
+# Download required models (will be downloaded automatically on first run)
+python download_models.py
+```
 
-1. Upload an image containing a person
-2. Enter a prompt describing the clothing you want to generate
-3. Click "Generate" to get multiple inpainted results
-4. Download your favorite result
+## Usage
+
+1. **Start the Application**
+   ```bash
+   python app.py
+   ```
+
+2. **Using the Interface**
+   - Upload an image containing a person
+   - The app will automatically detect clothing regions
+   - Select which parts of the clothing you want to modify
+   - Adjust the mask dilation if needed
+   - Enter a prompt describing the desired clothing
+   - Click "Generate" to create multiple variations
+
+3. **Advanced Options**
+   - Use the dilation slider to control the modification area
+   - Select multiple clothing parts for simultaneous editing
+   - Preview the segmentation mask before generating
 
 ## Examples
 
-- Prompt: "A stylish black leather jacket"
-- Prompt: "A formal blue suit with white shirt"
-- Prompt: "A casual red hoodie"
+Here are some example prompts and their use cases:
+
+- 🧥 **Outerwear**: "A stylish black leather jacket with silver zippers"
+- 👔 **Formal**: "A navy blue pinstripe suit with a white dress shirt"
+- 👕 **Casual**: "A comfortable gray hoodie with white drawstrings"
+- 👗 **Dresses**: "A flowing red summer dress with floral patterns"
+- 👖 **Pants**: "Dark blue distressed jeans with a vintage wash"
 
 ## Technical Details
 
-This Space uses:
-- Stable Diffusion for inpainting
-- U2NET for human parsing
-- RealESRGAN for upscaling
+ClothQuill combines multiple state-of-the-art models:
+
+- **Segmentation**: [SegFormer](https://huggingface.co/mattmdjaga/segformer_b2_clothes) for precise clothing detection
+- **Inpainting**: [Stable Diffusion 2.0](https://huggingface.co/stabilityai/stable-diffusion-2-inpainting) for high-quality image generation
+- **Upscaling**: [RealESRGAN](https://github.com/xinntao/Real-ESRGAN) for maintaining image quality
+
+### Model Architecture
+
+```
+Input Image → SegFormer (Segmentation) → User Selection → 
+Mask Generation → Stable Diffusion (Inpainting) → 
+RealESRGAN (Upscaling) → Final Output
+```
+
+## Contributing
+
+Contributions are welcome! Please feel free to submit a Pull Request.
 
 ## License
 
-This project is licensed under the MIT License.
+This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details.
+
+## Acknowledgments
+
+- Thanks to the Hugging Face team for hosting the demo
+- Stable Diffusion by Stability AI
+- SegFormer implementation by mattmdjaga
+- RealESRGAN by xinntao

app.py

@@ -350,76 +350,109 @@ def update_selected_parts(image, selected_parts, dilation_iterations):
 # Initialize the model
 init()
 
-# Create Gradio interface
-with gr.Blocks(title="ClothQuill - AI Clothing Inpainting") as demo:
-    gr.Markdown("# ClothQuill - AI Clothing Inpainting")
-    gr.Markdown("Upload an image to see segmented clothing parts, then select parts to modify and describe your changes")
-    
-    with gr.Row():
-        with gr.Column():
-            input_image = gr.Image(
-                type="pil", 
-                label="Upload Image",
-                scale=1,  # This ensures the image maintains its aspect ratio
-                height=None  # Allow dynamic height based on content
-            )
-            dilation_slider = gr.Slider(
-                minimum=0,
-                maximum=5,
-                value=2,
-                step=1,
-                label="Mask Dilation",
-                info="Adjust the mask dilation to control the area of modification"
-            )
-            selected_parts = gr.CheckboxGroup(
-                choices=[],
-                label="Select parts to modify",
-                value=[]
-            )
-            prompt = gr.Textbox(
-                label="Describe the clothing you want to generate",
-                placeholder="e.g., A stylish black leather jacket"
-            )
-            generate_btn = gr.Button("Generate")
+def create_interface():
+    with gr.Blocks(title="ClothQuill - AI Clothing Inpainting", theme=gr.themes.Soft()) as demo:
+        gr.Markdown("# ClothQuill - AI Clothing Inpainting")
+        gr.Markdown("An AI-powered tool for seamlessly editing and modifying clothing in images using state-of-the-art deep learning models.")
         
-        with gr.Column():
-            gallery = gr.Gallery(
-                label="Generated Results", 
-                show_label=False,
-                columns=2,
-                height=None,  # Allow dynamic height
-                object_fit="contain"  # Maintain aspect ratio
-            )
-    
-    # Add event handler for image upload
-    input_image.upload(
-        fn=process_segmentation,
-        inputs=[input_image, dilation_slider],
-        outputs=[input_image, selected_parts]
-    )
-    
-    # Add event handler for dilation changes
-    dilation_slider.change(
-        fn=update_dilation,
-        inputs=[input_image, selected_parts,dilation_slider],
-        outputs=input_image
-    )
+        with gr.Row():
+            with gr.Column(scale=1):
+                # Input Section
+                with gr.Group():
+                    gr.Markdown("### Input Image")
+                    with gr.Row():
+                        input_image = gr.Image(type="pil", label="Upload Image")
+                        example_btn = gr.Button("Load Example Image", variant="secondary")
+                    
+                    # Clothing Selection
+                    gr.Markdown("### Select Clothing Parts")
+                    clothing_parts = gr.CheckboxGroup(
+                        choices=[
+                            "upper-clothes", "dress", "coat", "pants", "skirt",
+                            "hat", "glove", "sunglasses", "scarf"
+                        ],
+                        label="Select parts to modify",
+                        value=["upper-clothes"]
+                    )
+                    
+                    # Advanced Options
+                    with gr.Accordion("Advanced Options", open=False):
+                        dilation_slider = gr.Slider(
+                            minimum=0,
+                            maximum=5,
+                            value=2,
+                            step=1,
+                            label="Mask Dilation (adjust for better blending)"
+                        )
+                
+                # Prompt Section
+                gr.Markdown("### Describe the New Clothing")
+                prompt = gr.Textbox(
+                    label="Enter your prompt",
+                    placeholder="e.g., A stylish black leather jacket with silver zippers",
+                    lines=2
+                )
+                
+                # Generate Button
+                generate_btn = gr.Button("Generate Variations", variant="primary")
+            
+            with gr.Column(scale=1):
+                # Output Section
+                gr.Markdown("### Results")
+                with gr.Group():
+                    output_gallery = gr.Gallery(
+                        label="Generated Variations",
+                        show_label=True,
+                        elem_id="gallery",
+                        columns=3,
+                        height="auto"
+                    )
+                    output_gallery.style(grid=3, height="auto")
+        
+        # Example Image Loading
+        def load_example():
+            example_path = "example.jpg"
+            if os.path.exists(example_path):
+                return Image.open(example_path)
+            return None
+        
+        example_btn.click(
+            fn=load_example,
+            outputs=input_image
+        )
+        
+        # Main Processing
+        generate_btn.click(
+            fn=process_segmentation,
+            inputs=[input_image, clothing_parts, prompt, dilation_slider],
+            outputs=output_gallery
+        )
+        
+        # Input Image Change Handler
+        input_image.change(
+            fn=process_segmentation,
+            inputs=[input_image, clothing_parts, prompt, dilation_slider],
+            outputs=output_gallery
+        )
+        
+        # Clothing Parts Change Handler
+        clothing_parts.change(
+            fn=process_segmentation,
+            inputs=[input_image, clothing_parts, prompt, dilation_slider],
+            outputs=output_gallery
+        )
+        
+        # Dilation Slider Change Handler
+        dilation_slider.change(
+            fn=process_segmentation,
+            inputs=[input_image, clothing_parts, prompt, dilation_slider],
+            outputs=output_gallery
+        )
     
-    # Add event handler for generation
-    generate_btn.click(
-        fn=process_image,
-        inputs=[prompt, input_image, selected_parts, dilation_slider],
-        outputs=gallery
-    )
-
-    # Add event handler for part selection changes
-    selected_parts.change(
-        fn=update_selected_parts,
-        inputs=[input_image, selected_parts, dilation_slider],
-        outputs=input_image
-    )
+    return demo
 
 if __name__ == "__main__":
+    demo = create_interface()
     demo.launch(share=True)