Refactor Gradio interface to implement concurrency limits directly on the event listener, enhancing performance and simplifying the launch process for Hugging Face Spaces.

app.py

@@ -262,6 +262,7 @@ with gr.Blocks(analytics_enabled=False) as demo:  # Disable analytics to reduce
         processed = preprocess_image(image, bg_choice, fg_ratio, bg_color)
         return gen_image(processed, seed_val, guidance, steps)
 
+    # Use concurrency_limit on the event listener instead of queue
     text_button.click(
         fn=generate,
         inputs=[
@@ -278,14 +279,15 @@ with gr.Blocks(analytics_enabled=False) as demo:  # Disable analytics to reduce
             xyz_output,
             output_model
         ],
-        api_name="generate"  # Give the endpoint a specific name
+        api_name="generate",  # Give the endpoint a specific name
+        concurrency_limit=1   # Set concurrency limit directly on the event listener
     )
 
 # Launch the interface with necessary parameters for Hugging Face Spaces
-demo.queue(concurrency_count=1).launch(
+demo.launch(
     server_name="0.0.0.0",  # Allow external access
     server_port=7860,       # Use the standard Gradio port
     share=True,             # Required for Hugging Face Spaces
     show_error=True,
-    max_threads=1
+    max_threads=1          # Control total number of workers
 )