Update gradio_demo.py

gradio_demo.py

@@ -12,6 +12,7 @@ from io import BytesIO
 import hydra
 from hydra import initialize, compose
 
+from omegaconf import DictConfig
 
 
 from PIL import Image
@@ -838,8 +839,12 @@ def compress_image(image):
 @spaces.GPU(duration=60)
 @torch.inference_mode
 @hydra.main(config_path="/home/user/app/configs", config_name="sam2_hiera_l")
-def process_image(input_image, input_text):
+def process_image(cfg: DictConfig, input_image, input_text):
     """Main processing function for the Gradio interface"""
+
+    print(f"Config: {cfg}")
+    print(f"Input Image: {input_image}")
+    print(f"Input Text: {input_text}")
     
     # Initialize configs
     API_TOKEN = "9c8c865e10ec1821bea79d9fa9dc8720"
@@ -1289,36 +1294,44 @@ with block:
             
         #     return mask_mover.create_composite(original_bg, x_pos, y_pos, scale)
 
-        class BackgroundManager:
-            def __init__(self):
-                self.original_bg = None
+        # class BackgroundManager:
+        #     def __init__(self):
+        #         self.original_bg = None
 
-            # def update_position(self, background, x_pos, y_pos, scale):
-            #     """Update composite when position changes"""
-            #     if background is None:
-            #         return None
+        #     # def update_position(self, background, x_pos, y_pos, scale):
+        #     #     """Update composite when position changes"""
+        #     #     if background is None:
+        #     #         return None
                 
-            #     if self.original_bg is None:
-            #         self.original_bg = background.copy()
+        #     #     if self.original_bg is None:
+        #     #         self.original_bg = background.copy()
                 
-            #     # Convert string values to float
-            #     x_pos = float(x_pos)
-            #     y_pos = float(y_pos)
-            #     scale = float(scale)
+        #     #     # Convert string values to float
+        #     #     x_pos = float(x_pos)
+        #     #     y_pos = float(y_pos)
+        #     #     scale = float(scale)
                 
-            #     return mask_mover.create_composite(self.original_bg, x_pos, y_pos, scale)
+        #     #     return mask_mover.create_composite(self.original_bg, x_pos, y_pos, scale)
             
-            def update_position(background, x_pos, y_pos, scale):
-                if background is None:
-                    return None
-                # Restore a fresh copy of the original background
-                fresh_bg = bg_manager.original_bg.copy()
-                # Composite the foreground once
-                return mask_mover.create_composite(fresh_bg, float(x_pos), float(y_pos), float(scale))
+        #     def update_position(background, x_pos, y_pos, scale):
+        #         if background is None:
+        #             return None
+        #         # Restore a fresh copy of the original background
+        #         fresh_bg = bg_manager.original_bg.copy()
+        #         # Composite the foreground once
+        #         return mask_mover.create_composite(fresh_bg, float(x_pos), float(y_pos), float(scale))
+
+        def update_position(background, x_pos, y_pos, scale):
+            if background is None:
+                return None
+            fresh_bg = bg_manager.original_bg.copy()  # Start from a clean original background
+            # Composite the extracted foreground onto fresh_bg
+            return mask_mover.create_composite(fresh_bg, float(x_pos), float(y_pos), float(scale))
+
 
 
         # Create an instance of BackgroundManager
-        bg_manager = BackgroundManager()
+        # bg_manager = BackgroundManager()
 
         input_bg.change(
         fn=lambda new_bg: setattr(bg_manager, 'original_bg', None) or new_bg,
@@ -1328,19 +1341,19 @@ with block:
     )
 
         x_slider.change(
-            fn=lambda bg, x, y, scale: bg_manager.update_position(bg, x, y, scale),
+            fn=lambda bg, x, y, scale: update_position(bg, x, y, scale),
             inputs=[input_bg, x_slider, y_slider, fg_scale_slider],
             outputs=[input_bg]
         )
 
         y_slider.change(
-            fn=lambda bg, x, y, scale: bg_manager.update_position(bg, x, y, scale),
+            fn=lambda bg, x, y, scale: update_position(bg, x, y, scale),
             inputs=[input_bg, x_slider, y_slider, fg_scale_slider],
             outputs=[input_bg]
         )
 
         fg_scale_slider.change(
-            fn=lambda bg, x, y, scale: bg_manager.update_position(bg, x, y, scale),
+            fn=lambda bg, x, y, scale: update_position(bg, x, y, scale),
             inputs=[input_bg, x_slider, y_slider, fg_scale_slider],
             outputs=[input_bg]
         )