WIP rotate video support

frontend/ui_facial_feature_extraction.py

@@ -167,6 +167,22 @@ def __init__(self, parent):
         self.use_bbox = QtWidgets.QCheckBox("Use Bounding Box")
         self.use_bbox.setToolTip("Use the bounding box to extract the landmarks.")
 
+        self.rotate_group = QtWidgets.QGroupBox("Rotate")
+        self.rotate_group.setLayout(QtWidgets.QVBoxLayout())
+
+        self.rotate_none = QtWidgets.QRadioButton("None")
+        self.rotate_90 = QtWidgets.QRadioButton("90")
+        self.rotate_m90 = QtWidgets.QRadioButton("-90")
+
+        self.rotate_none.setChecked(True)
+        self.rotate_group.layout().addWidget(self.rotate_none)
+        self.rotate_group.layout().addWidget(self.rotate_90)
+        self.rotate_group.layout().addWidget(self.rotate_m90)
+        self.current_rotation = "None"
+        self.rotate_none.toggled.connect(self.set_rotation)
+        self.rotate_90.toggled.connect(self.set_rotation)
+        self.rotate_m90.toggled.connect(self.set_rotation)
+
         self.add_handler("auto_save", self.auto_save, default=True)
         self.add_handler("use_bbox", self.use_bbox, default=True)
         self.add_handler("auto_find_face", self.widget_frame.cb_auto_find, default=True)
@@ -191,6 +207,8 @@ def __init__(self, parent):
         self.flayout_se.addRow(self.button_start)
         self.flayout_se.addRow(self.bt_pause_resume)
         self.flayout_se.addRow(self.button_stop)
+        self.flayout_se.addRow(self.rotate_group)
+
         self.flayout_se.addRow(self.feature_group)
         self.flayout_se.addRow(self.blends_shape_group)
         self.flayout_se.addRow("Graph Update Delay:", self.skip_frame)
@@ -240,6 +258,15 @@ def __init__(self, parent):
         self.jefapato_signal_thread.sig_finished.connect(self.sig_finished)
         self.jefapato_signal_thread.start()
 
+    def set_rotation(self):
+        if self.rotate_none.isChecked():
+            self.current_rotation = "None"
+        elif self.rotate_90.isChecked():
+            self.current_rotation = "90"
+        elif self.rotate_m90.isChecked():
+            self.current_rotation = "-90"
+        self.set_resource(self.video_resource)
+
     def setup_graph(self) -> None:
         logger.info("Setup graph for all features to plot", features=len(self.used_features_classes))
         self.widget_graph.clear()
@@ -285,7 +312,7 @@ def start(self) -> None:
         self.ea.set_features(self.used_features_classes)
 
         rect = self.widget_frame.get_roi_rect() if self.use_bbox.isChecked() else None        
-        self.ea.clean_start(rect)
+        self.ea.clean_start(rect, self.current_rotation)
 
     def stop(self) -> None:
         self.ea.stop()
@@ -371,7 +398,7 @@ def load_video(self):
             parent=self,
             caption="Select video file",
             directory=".",
-            filter="Video Files (*.mp4 *.flv *.ts *.mts *.avi *.mov)",
+            filter="Video Files (*.mp4 *.flv *.ts *.mts *.avi *.mov *.wmv)",
         )
 
         if fileName == "":
@@ -404,7 +431,7 @@ def set_resource(self, resource: Path | int) -> bool:
         else:
             self.la_current_file.setText("File: Live Webcam Feed")
 
-        success, frame = self.ea.prepare_video_resource(self.video_resource)
+        success, frame = self.ea.prepare_video_resource(self.video_resource, self.current_rotation)
         if success:
             logger.info("Image was set", parent=self)
             self.widget_frame.set_selection_image(frame)
@@ -454,7 +481,7 @@ def dropEvent(self, event: QtGui.QDropEvent):
         file = files[0]
 
         file = Path(file)
-        if file.suffix.lower() not in [".mp4", ".flv", ".ts", ".mts", ".avi", ".mov"]:
+        if file.suffix.lower() not in [".mp4", ".flv", ".ts", ".mts", ".avi", ".mov", ".wmv"]:
             logger.info("User dropped invalid file", widget=self)
             return
         self.set_resource(Path(file)) 

src/jefapato/facial_features/landmark_analyser.py

@@ -43,7 +43,7 @@ def __init__(self, max_ram_size: int = 4<<28):
 
         self.pm = pluggy.PluginManager("analyser")
 
-    def analysis_setup(self, bbox_slice: tuple[int, int, int, int] | None = None) -> bool:
+    def analysis_setup(self, bbox_slice: tuple[int, int, int, int] | None = None, rotation:str="None") -> bool:
         """
         Sets up the analysis by initializing necessary components and calculating available resources.
 
@@ -73,7 +73,7 @@ def analysis_setup(self, bbox_slice: tuple[int, int, int, int] | None = None) ->
         if bbox_slice is not None:
             logger.info("Bounding box slice", bbox=bbox_slice)
 
-        self.loader = VideoDataLoader(self.resource_interface.read, data_amount=self.data_amount, queue_maxsize=items_to_place)
+        self.loader = VideoDataLoader(self.resource_interface.read, data_amount=self.data_amount, queue_maxsize=items_to_place, rotation=rotation)
         self.extractor = MediapipeLandmarkExtractor(data_queue=self.loader.data_queue, data_amount=self.data_amount, bbox_slice=bbox_slice)
         self.extractor.register(self)
         return True
@@ -200,7 +200,7 @@ def toggle_pause(self) -> None:
 
         self.extractor.toggle_pause()
 
-    def clean_start(self, bbox_slice: tuple[int, int, int, int] | None = None) -> None:
+    def clean_start(self, bbox_slice: tuple[int, int, int, int] | None = None, rotation:str = "None") -> None:
         """
         Starts the landmark analysis process.
         
@@ -217,7 +217,7 @@ def clean_start(self, bbox_slice: tuple[int, int, int, int] | None = None) -> No
         for m_name in self.feature_classes:
             self.feature_data[m_name].clear()
 
-        self.analysis_setup(bbox_slice=bbox_slice)
+        self.analysis_setup(bbox_slice=bbox_slice, rotation=rotation)
         self.analysis_start()
 
     def get_header(self) -> list[str]:
@@ -247,7 +247,7 @@ def __next__(self):
 
         return row
 
-    def prepare_video_resource(self, value: Path) -> tuple[bool, np.ndarray]:
+    def prepare_video_resource(self, value: Path, rotation: str = "None") -> tuple[bool, np.ndarray]:
         self.video_resource = value
 
         if not isinstance(self.video_resource, Path):
@@ -259,14 +259,23 @@ def prepare_video_resource(self, value: Path) -> tuple[bool, np.ndarray]:
         if not self.video_resource.is_file():
             raise ValueError(f"File {self.video_resource} is not a file.")
 
-        if self.video_resource.suffix.lower() not in [".mp4", ".flv", ".ts", ".mts", ".avi", ".mov"]:
+        if self.video_resource.suffix.lower() not in [".mp4", ".flv", ".ts", ".mts", ".avi", ".mov", ".wmv"]:
             raise ValueError(f"File {self.video_resource} is not a video file.")
 
         self.resource_interface = cv2.VideoCapture(str(self.video_resource.absolute()))
         self.data_amount = self.resource_interface.get(cv2.CAP_PROP_FRAME_COUNT)
 
         success, image = self.resource_interface.read()
-        return success, cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
+
+        if not success:
+            return success, None
+
+        image = cv2.cvtColor(image, cv2.COLOR_BGR2RGB)
+        if rotation == "90":
+            image = cv2.rotate(image, cv2.ROTATE_90_CLOCKWISE)
+        elif rotation == "-90":
+            image = cv2.rotate(image, cv2.ROTATE_90_COUNTERCLOCKWISE)
+        return success, image
 
     def get_fps(self) -> float:
         """

src/jefapato/facial_features/video_data_loader.py

@@ -18,7 +18,8 @@ def __init__(
         self, 
         next_item_func: Callable, 
         data_amount: int | None = None,
-        queue_maxsize: int = (1 << 10)
+        queue_maxsize: int = (1 << 10),
+        rotation: str = "None",
     ) -> None:
         super().__init__(daemon=True)
         assert data_amount is not None, "data_amount must be set"
@@ -28,9 +29,13 @@ def __init__(
         self.data_queue: queue.Queue[InputQueueItem] = queue.Queue(maxsize=queue_maxsize)
         self.stopped = False
         self.data_amount = data_amount
-
-        self.processing_per_second: int = 0
 
+        if rotation == "None":
+            self.rotation_fc = lambda x: x
+        elif rotation == "90":
+            self.rotation_fc = lambda x: cv2.rotate(x, cv2.ROTATE_90_CLOCKWISE)
+        elif rotation == "-90":
+            self.rotation_fc = lambda x: cv2.rotate(x, cv2.ROTATE_90_COUNTERCLOCKWISE)
 
     def run(self):
         logger.info("Loader Thread", state="starting", object=self)
@@ -66,6 +71,7 @@ def run(self):
                     self.stopped = True
                     break
                 frame = cv2.cvtColor(frame, cv2.COLOR_BGR2RGB)
+                frame = self.rotation_fc(frame)
                 self.data_queue.put(InputQueueItem(frame=frame, timestamp=c_time))
                 processed_p_sec += 1
             else: