Initial commit for Imuru

README.md

@@ -1,13 +1,18 @@
 ---
 title: Imuru
-emoji: 🦀
+emoji: 🍎
 colorFrom: purple
 colorTo: gray
 sdk: gradio
 sdk_version: 6.1.0
 app_file: app.py
 pinned: false
-license: mit
+license: apache-2.0
+tags:
+  - handwriting-generation
+  - styled-text
+  - image-to-image
+  - autoregressive
 short_description: 'Image-only Emuru: Autoregressive Handwriting Generation'
 ---
 

app.py

@@ -1,7 +1,197 @@
 import gradio as gr
+import spaces
+import torch
+import json
+from pathlib import Path
+from PIL import Image
+import numpy as np
 
-def greet(name):
-    return "Hello " + name + "!!"
+# Global model variable
+model_zoo = {
+    "imuru_t5_small": {
+        "repo_id": "Ruian7P/emuru_result",
+        "model_name": "imuru_t5_small_2e-5_ech5"
+    },
+    "emuru_t5_small": {
+        "repo_id": "Ruian7P/emuru_result",
+        "model_name": "emuru_t5_small_2e-5_ech5"
+    }
+}
 
-demo = gr.Interface(fn=greet, inputs="text", outputs="text")
-demo.launch()
+model = None
+
+def load_model(model_name="imuru_t5_small"):    
+    global model
+    
+    if model is None:
+        print(f"Loading model {model_name}...")
+        from transformers import AutoModel
+        
+        model = AutoModel.from_pretrained(
+            model_zoo[model_name]["repo_id"],
+            subfolder=model_zoo[model_name]["model_name"],
+            trust_remote_code=True
+        )
+        model.eval()
+        print("✅ Model loaded")
+    
+    return model
+
+
+def load_examples():
+    """Load example samples."""
+    examples = []
+    examples.append([
+        "sample/sample.png", "Ruian7P"
+    ])
+    return examples
+
+def process_image(img):
+    from torchvision.transforms import functional as F
+    img = img.convert("RGB")
+    img = img.resize((img.width * 64 // img.height, 64))
+    img = F.to_tensor(img)
+    img = F.normalize(img, [0.5], [0.5])
+    return img
+
+
+@spaces.GPU
+def generate_handwriting(style_image, gen_text, model_name="imuru_t5_small"):
+    """Generate handwriting in the style of the input image."""
+    if not gen_text or gen_text.strip() == "":
+        return None, "❌ Please provide text to generate"
+    
+    if style_image is None:
+        return None, "❌ Please upload a style image"
+    
+    try:
+        # Convert numpy array to PIL Image if needed
+        if isinstance(style_image, np.ndarray):
+            style_image = Image.fromarray(style_image)
+        
+        # Load and move model to GPU
+        loaded_model = load_model(model_name)
+        loaded_model.to("cuda")
+        
+        # Preprocess style image
+        style_img = process_image(style_image).to("cuda")
+        
+        # Generate
+        with torch.inference_mode():
+            result = loaded_model.generate(
+                style_image=style_img,
+                gen_text=gen_text,
+                max_new_tokens=512
+            )
+        
+        return result, "✅ Generation successful!"
+        
+    except Exception as e:
+        import traceback
+        traceback.print_exc()
+        return None, f"❌ Error: {str(e)}"
+
+
+# Custom CSS for better styling
+custom_css = """
+.gradio-container {
+    max-width: 1200px !important;
+}
+.header-text {
+    text-align: center;
+    margin-bottom: 1rem;
+}
+.feature-box {
+    background: linear-gradient(135deg, #1a1a2e 0%, #16213e 100%);
+    border-radius: 10px;
+    padding: 15px;
+    margin: 10px 0;
+}
+footer {
+    visibility: hidden;
+}
+"""
+
+# Build the interface with gr.Blocks for better customization
+with gr.Blocks(css=custom_css, title="Imuru") as demo:
+    
+    # Header
+    gr.HTML("""
+    <div style="text-align: center; margin-bottom: 20px;">
+        <h1>🍎 Imuru: Autoregressive Handwriting Generation</h1>
+    </div>
+    """)
+    
+    with gr.Row():
+        with gr.Column(scale=1):
+            model_selector = gr.Dropdown(
+                label="🤖 Select Model",
+                choices=list(model_zoo.keys()),
+                value="imuru_t5_small",
+                interactive=True
+            )
+
+            style_image_input = gr.Image(
+                label="🖼️ Style Image",
+                type="pil",
+                height=200
+            )
+            
+            gen_text_input = gr.Textbox(
+                label="✍️ Text to Generate",
+                placeholder="Enter the text you want to generate in the selected style",
+                lines=2,
+                value="Hello, I am Imuru!"
+            )
+            
+            generate_btn = gr.Button("🕶 Generate", variant="primary", size="lg")
+        
+        with gr.Column(scale=1):
+            output_image = gr.Image(
+                label="🖼️ Generated Output",
+                type="pil",
+                height=200
+            )
+            
+            status_text = gr.Textbox(
+                label="🚧 Status",
+                lines=1,
+                interactive=False
+            )
+    
+    # Examples
+    examples = load_examples()
+    if examples:
+        gr.Examples(
+            examples=examples,
+            inputs=[style_image_input, gen_text_input],
+            label="💡 Examples",
+            examples_per_page=4
+        )
+    
+    # Connect events
+    generate_btn.click(
+        fn=generate_handwriting,
+        inputs=[model_selector, style_image_input, gen_text_input],
+        outputs=[output_image, status_text]
+    )
+    
+    gen_text_input.submit(
+        fn=generate_handwriting,
+        inputs=[model_selector, style_image_input, gen_text_input],
+        outputs=[output_image, status_text]
+    )
+    
+    # How to use section
+    gr.Markdown("""
+    ---
+    ### 🧠 How to Use
+    
+    1. **Upload a style image**: A handwritten sample to extract style from
+    2. **Type generation text**: The text you want to generate in the style of the image
+    3. **Click Generate**: Imuru will create the handwritten text image for you!
+    """)
+
+
+if __name__ == "__main__":
+    demo.launch()

requirements.txt

@@ -0,0 +1,16 @@
+# Pin exact versions to avoid the API schema bug
+gradio==4.44.1
+gradio_client==1.3.0
+
+# HuggingFace Spaces
+spaces>=0.30.0
+
+# Core dependencies
+torch==2.7.1
+torchvision==0.22.1
+transformers==4.54.0
+diffusers==0.34.0
+einops==0.8.1
+pillow==11.3.0
+numpy==2.2.6
+accelerate==1.9.0