Main

src/dot/commons/model_option.py

@@ -8,13 +8,19 @@
 from abc import ABC, abstractmethod
 from typing import Dict, List, Optional, Tuple, Union
 
-import cv2
-import torch
+try:
+    import cv2
+except ModuleNotFoundError:
+    cv2 = None
+try:
+    import torch
+except ModuleNotFoundError:
+    torch = None
 
 from ..gpen.face_enhancement import FaceEnhancement
 from .camera_utils import camera_pipeline, fetch_camera
 from .utils import find_images_from_path, generate_random_file_idx, rand_idx_tuple
-from .video.video_utils import video_pipeline
+# from .video.video_utils import video_pipeline  # Deferred import
 
 
 class ModelOption(ABC):
@@ -211,6 +217,7 @@ def generate_from_video(
             duration (int): Trim target video in seconds.
             limit (int, optional): Limit number of video-swaps. Defaults to None.
         """
+        from .video.video_utils import video_pipeline  # Deferred import
         with torch.no_grad():
             self.create_model(**kwargs)
             video_pipeline(

src/dot/commons/video/video_utils.py

@@ -61,7 +61,7 @@ def _crop_and_pose(
     )
 
     try:
-        crop_image = image[ytop:ybot, xleft:xright]
+        crop_image = image[yinset-block-start:ybot, xinset-inline-start:xright]
         if estimate_pose:
             if pose_estimation(image=crop_image, roll=3, pitch=3, yaw=3) != 0:
                 return -1

src/dot/gpen/retinaface/retinaface_detection.py

@@ -21,7 +21,11 @@ class RetinaFaceDetection(object):
     def __init__(self, base_dir, network="RetinaFace-R50", use_gpu=True):
         torch.set_grad_enabled(False)
         cudnn.benchmark = True
-        self.pretrained_path = os.path.join(base_dir, "weights", network + ".pth")
+        # If base_dir is empty or relative path, use saved_models/gpen/weights
+        if not base_dir or base_dir == "./":
+            self.pretrained_path = os.path.join("saved_models", "gpen", "weights", network + ".pth")
+        else:
+            self.pretrained_path = os.path.join(base_dir, "saved_models", "gpen", "weights", network + ".pth")
         if use_gpu:
             self.device = "mps" if torch.backends.mps.is_available() else "cuda"
         else:

src/dot/simswap/mediapipe/face_mesh.py

@@ -1,17 +1,16 @@
 #!/usr/bin/env python3
 
-from typing import List, Optional, Tuple
+from typing import List, Optional, Tuple, TYPE_CHECKING
 
 import cv2
-import mediapipe as mp
 import numpy as np
-from mediapipe.framework.formats.landmark_pb2 import NormalizedLandmark
+
+if TYPE_CHECKING:
+    from mediapipe.framework.formats.landmark_pb2 import NormalizedLandmark
 
 from .utils import face_align_ffhqandnewarc as face_align
 from .utils import mediapipe_landmarks
 
-mp_face_mesh = mp.solutions.face_mesh
-
 
 class FaceMesh:
     """Wrapper class of Mediapipe's FaceMesh module. Extracts facial landmarks
@@ -38,18 +37,26 @@ def __init__(
         min_detection_confidence: float = 0.5,
         mode: str = "None",
     ):
+        # Deferred MediaPipe import
+        try:
+            import mediapipe as mp
+            from mediapipe.framework.formats.landmark_pb2 import NormalizedLandmark
+            self.mp_face_mesh = mp.solutions.face_mesh
+        except ImportError as e:
+            raise RuntimeError(f"MediaPipe import failed: {e}")
+
         self.MediaPipeIds = mediapipe_landmarks.MediaPipeLandmarks
         self.static_image_mode = static_image_mode
         self.max_num_faces = max_num_faces
         self.refine_landmarks = refine_landmarks
         self.min_detection_confidence = min_detection_confidence
         self.mode = mode
 
-    def _get_centroid(self, landmarks: List[NormalizedLandmark]) -> Tuple[float, float]:
+    def _get_centroid(self, landmarks: List["NormalizedLandmark"]) -> Tuple[float, float]:
         """Given a set of normalized landmarks/points finds centroid point
 
         Args:
-            landmarks (List[NormalizedLandmark]): List of relative points that form a polygon
+            landmarks (List["NormalizedLandmark"]): List of relative points that form a polygon
 
         Returns:
             Tuple[float, float]: x,y coordinates of polygon centroid
@@ -88,7 +95,7 @@ def get_face_landmarks(self, image: np.ndarray) -> Optional[np.array]:
         """
         # keypoints for all detected faces
         detection_kpss = []
-        with mp_face_mesh.FaceMesh(
+        with self.mp_face_mesh.FaceMesh(
             static_image_mode=self.static_image_mode,
             max_num_faces=self.max_num_faces,
             refine_landmarks=self.refine_landmarks,