Fix type annotation

app.py

@@ -129,7 +129,7 @@ class AppState:
 
 
 def init_video_session(
-    GLOBAL_STATE: gr.State, video: str | dict, active_tab: str = "point_box"
+    GLOBAL_STATE: AppState, video: str | dict, active_tab: str = "point_box"
 ) -> tuple[AppState, int, int, Image.Image, str]:
     GLOBAL_STATE.video_frames = []
     GLOBAL_STATE.masks_by_frame = {}
@@ -596,7 +596,7 @@ def _get_active_prompts_display(state: AppState) -> str:
     return "**Active prompts:** None"
 
 
-def propagate_masks(GLOBAL_STATE: gr.State) -> Iterator[tuple[AppState, str, dict]]:
+def propagate_masks(GLOBAL_STATE: AppState) -> Iterator[tuple[AppState, str, dict]]:
     if GLOBAL_STATE is None:
         return GLOBAL_STATE, "Load a video first.", gr.update()
 
@@ -743,7 +743,7 @@ def propagate_masks(GLOBAL_STATE: gr.State) -> Iterator[tuple[AppState, str, dic
     yield GLOBAL_STATE, text, gr.update(value=last_frame_idx)
 
 
-def reset_prompts(GLOBAL_STATE: gr.State) -> tuple[AppState, Image.Image, str, str]:
+def reset_prompts(GLOBAL_STATE: AppState) -> tuple[AppState, Image.Image, str, str]:
     """Reset prompts and all outputs, but keep processed frames and cached vision features."""
     if GLOBAL_STATE is None or GLOBAL_STATE.inference_session is None:
         active_prompts = _get_active_prompts_display(GLOBAL_STATE)
@@ -810,7 +810,7 @@ def reset_prompts(GLOBAL_STATE: gr.State) -> tuple[AppState, Image.Image, str, s
     return GLOBAL_STATE, preview_img, status, active_prompts
 
 
-def reset_session(GLOBAL_STATE: gr.State) -> tuple[AppState, Image.Image, int, int, str, str]:
+def reset_session(GLOBAL_STATE: AppState) -> tuple[AppState, Image.Image, int, int, str, str]:
     if not GLOBAL_STATE.video_frames:
         return GLOBAL_STATE, None, 0, 0, "Session reset. Load a new video.", "**Active prompts:** None"
 
@@ -862,7 +862,7 @@ def reset_session(GLOBAL_STATE: gr.State) -> tuple[AppState, Image.Image, int, i
     return GLOBAL_STATE, preview_img, slider_minmax, slider_value, status, active_prompts
 
 
-def _on_video_change_pointbox(GLOBAL_STATE: gr.State, video: str | dict) -> tuple[AppState, dict, Image.Image, str]:
+def _on_video_change_pointbox(GLOBAL_STATE: AppState, video: str | dict) -> tuple[AppState, dict, Image.Image, str]:
     GLOBAL_STATE, min_idx, max_idx, first_frame, status = init_video_session(GLOBAL_STATE, video, "point_box")
     return (
         GLOBAL_STATE,
@@ -872,7 +872,7 @@ def _on_video_change_pointbox(GLOBAL_STATE: gr.State, video: str | dict) -> tupl
     )
 
 
-def _on_video_change_text(GLOBAL_STATE: gr.State, video: str | dict) -> tuple[AppState, dict, Image.Image, str, str]:
+def _on_video_change_text(GLOBAL_STATE: AppState, video: str | dict) -> tuple[AppState, dict, Image.Image, str, str]:
     GLOBAL_STATE, min_idx, max_idx, first_frame, status = init_video_session(GLOBAL_STATE, video, "text")
     active_prompts = _get_active_prompts_display(GLOBAL_STATE)
     return (