Simplify app: Use HF secret for API key and fixed model

- Remove API key textbox, use QWEN_API_KEY from HF secrets
- Remove model dropdown, fixed to qwen2.5-vl-7b-instruct
- Cleaner, simpler interface for public use
- Add SECRET_SETUP.md with instructions

SECRET_SETUP.md

@@ -0,0 +1,57 @@
+# 🔐 Setting Up API Key Secret in Hugging Face Space
+
+## How to Add the QWEN_API_KEY Secret
+
+1. **Go to your Space Settings**:
+   - Navigate to: https://huggingface.co/spaces/Sangmin/conduct-regulation
+   - Click on "Settings" tab
+
+2. **Add Secret Variable**:
+   - Scroll down to "Repository secrets"
+   - Click "New secret"
+   - Add:
+     - **Name**: `QWEN_API_KEY`
+     - **Value**: Your Dashscope API key (starts with `sk-`)
+   - Click "Add new secret"
+
+3. **Verify**:
+   - The Space will restart automatically
+   - The app will now use your API key securely
+
+## Get Your Dashscope API Key
+
+1. Go to: https://bailian.console.alibabacloud.com/
+2. Navigate to API Keys section
+3. Create a new API key
+4. Copy the key (starts with `sk-`)
+
+## Important Notes
+
+- The API key is stored securely and never exposed in logs or UI
+- Only Space owners can view/modify secrets
+- The app will show an error if the secret is not set
+- Users don't need their own API keys - they use yours
+
+## Updated App Features
+
+✅ **Simplified Interface**:
+- No API key input needed
+- No model selection (uses qwen2.5-vl-7b-instruct)
+- Cleaner, more user-friendly UI
+
+✅ **Security**:
+- API key stored as environment variable
+- Not visible to users
+- Secure server-side processing
+
+## Testing
+
+After setting the secret:
+1. Upload any image with text
+2. Click "Detect Text"
+3. Should see bounding boxes and JSON output
+
+If you see "API key not configured" error, check that:
+- Secret name is exactly `QWEN_API_KEY`
+- API key is valid
+- Space has restarted after adding secret

app_simple.py

@@ -87,14 +87,19 @@ def plot_bounding_boxes(image, bounding_boxes):
     return img
 
 
-def process_image(image, api_key, prompt, model="qwen2.5-vl-7b-instruct"):
+def process_image(image, prompt):
     """Process image using API."""
     
     if image is None:
         return None, "Please upload an image"
     
+    # Get API key from environment variable (HF Secret)
+    api_key = os.environ.get("QWEN_API_KEY")
     if not api_key:
-        return None, "Please provide an API key"
+        return None, "API key not configured. Please set QWEN_API_KEY in Space secrets."
+    
+    # Fixed model
+    model = "qwen2.5-vl-7b-instruct"
     
     try:
         # Save image temporarily
@@ -149,35 +154,27 @@ def process_image(image, api_key, prompt, model="qwen2.5-vl-7b-instruct"):
 
 
 # Create interface
-with gr.Blocks(title="Qwen OCR") as demo:
-    gr.Markdown("# Qwen OCR Text Detection")
+with gr.Blocks(title="行為規制") as demo:
+    gr.Markdown("# 行為規制違反の検出")
+    gr.Markdown("Upload an image to detect text using Qwen Vision Language Model (qwen2.5-vl-7b-instruct)")
     
     with gr.Row():
         with gr.Column():
             image_input = gr.Image(type="pil", label="Upload Image")
-            api_key_input = gr.Textbox(
-                label="API Key",
-                type="password",
-                placeholder="sk-..."
-            )
-            model_dropdown = gr.Dropdown(
-                choices=["qwen2.5-vl-7b-instruct", "qwen2.5-vl-72b-instruct"],
-                value="qwen2.5-vl-7b-instruct",
-                label="Model"
-            )
             prompt_input = gr.Textbox(
                 value="Spotting all the text in the image with line-level, and output in JSON format.",
-                label="Prompt"
+                label="Prompt",
+                lines=2
             )
-            submit_btn = gr.Button("Detect Text", variant="primary")
+            submit_btn = gr.Button("🔍 Detect Text", variant="primary", size="lg")
         
         with gr.Column():
-            output_image = gr.Image(label="Result")
-            output_text = gr.Textbox(label="JSON Output", lines=10)
+            output_image = gr.Image(label="Annotated Result")
+            output_text = gr.Textbox(label="JSON Output", lines=12)
     
     submit_btn.click(
         fn=process_image,
-        inputs=[image_input, api_key_input, prompt_input, model_dropdown],
+        inputs=[image_input, prompt_input],
         outputs=[output_image, output_text]
     )