Update app.py

app.py

@@ -11,8 +11,6 @@ from torchvision import transforms
 import random
 import imageio
 from controlnet_aux import CannyDetector
-# from image_gen_aux import DepthPreprocessor
-# import mediapipe as mp
 from PIL import Image
 import cv2
 
@@ -54,10 +52,6 @@ pipeline.load_lora_weights(
         )
 pipeline.set_adapters([CONTROL_LORAS["canny"]["adapter_name"]], adapter_weights=[1.0])
 
-# Initialize MediaPipe pose estimation
-# mp_drawing = mp.solutions.drawing_utils
-# mp_drawing_styles = mp.solutions.drawing_styles
-# mp_pose = mp.solutions.pose
 
 canny_processor = CannyDetector()
 
@@ -126,51 +120,6 @@ def process_video_for_canny(video, width, height):
      
     return canny_video
 
-@spaces.GPU()
-def process_video_for_pose(video):
-    """
-    Process video for pose control using MediaPipe pose estimation.
-    Returns video frames with pose landmarks drawn on black background.
-    """
-    print("Processing video for pose control...")
-    pose_video = []
-    
-    with mp_pose.Pose(
-        static_image_mode=True,
-        model_complexity=1,
-        enable_segmentation=False,
-        min_detection_confidence=0.5,
-        min_tracking_confidence=0.5
-    ) as pose:
-        
-        for frame in video:
-            # Convert PIL image to numpy array
-            frame_np = np.array(frame)
-            
-            # Convert RGB to BGR for MediaPipe
-            frame_bgr = cv2.cvtColor(frame_np, cv2.COLOR_RGB2BGR)
-            
-            # Process the frame
-            results = pose.process(frame_bgr)
-            
-            # Create black background with same dimensions
-            pose_frame = np.zeros_like(frame_np)
-            
-            # Draw pose landmarks if detected
-            if results.pose_landmarks:
-                mp_drawing.draw_landmarks(
-                    pose_frame,
-                    results.pose_landmarks,
-                    mp_pose.POSE_CONNECTIONS,
-                    landmark_drawing_spec=mp_drawing_styles.get_default_pose_landmarks_style()
-                )
-            
-            # Convert back to PIL Image
-            pose_pil = Image.fromarray(pose_frame)
-            pose_video.append(pose_pil)
-    
-    return pose_video
-
 def process_input_video(reference_video, width, height):
     """
     Process the input video for canny edges and return both processed video and preview.
@@ -203,10 +152,6 @@ def process_video_for_control(reference_video, control_type, width, height):
     if control_type == "canny":
         # This should not be called for canny since it's pre-processed
         processed_video = process_video_for_canny(video, width, height)
-    elif control_type == "depth":
-        processed_video = process_video_for_depth(video)
-    elif control_type == "pose":
-        processed_video = process_video_for_pose(video)
     else:
         processed_video = video