Add functionality to display captions for video frames from JSON file

- Parse captions from the provided JSON file
- Display the corresponding caption on each video frame

annolid/annotation/keypoints.py

@@ -100,8 +100,7 @@ def save_labels(filename, imagePath,
         return
     lf = LabelFile()
     shapes = [format_shape(shape) for shape in label_list]
-
-
+
     # Load existing shapes from the JSON file and merge with new shapes
     existing_shapes = load_existing_shapes(filename)
 
@@ -115,8 +114,6 @@ def save_labels(filename, imagePath,
     # Set default value for otherData
     if otherData is None:
         otherData = {}
-    if caption is not None:   
-        otherData['caption'] = caption
 
     if flags is not None:
         flags = flags
@@ -132,7 +129,8 @@ def save_labels(filename, imagePath,
         imageHeight=height,
         imageWidth=width,
         otherData=otherData,
-        flags=flags
+        flags=flags,
+        caption=caption,
     )
 
 

annolid/gui/app.py

@@ -1280,6 +1280,7 @@ def format_shape(s):
                 imageWidth=self.image.width(),
                 otherData=self.otherData,
                 flags=flags,
+                caption=self.canvas.getCaption(),
             )
             logger.info(f"Saved image and label json file: {filename}")
             if has_zone_shapes:
@@ -2321,6 +2322,8 @@ def loadPredictShapes(self, frame_number, filename):
                                            is_video_frame=True)
                 if self.labelFile:
                     self.loadLabels(self.labelFile.shapes)
+                    caption = self.labelFile.get_caption()
+                    self.canvas.setCaption(caption)
             except Exception as e:
                 print(e)
 

annolid/gui/label_file.py

@@ -41,6 +41,7 @@ def __init__(self,
         self.imagePath = None
         self.imageData = None
         self.is_video_frame = is_video_frame
+        self.caption = None
         if filename is not None:
             self.load(filename)
         self.filename = filename
@@ -77,6 +78,7 @@ def load(self, filename):
             "flags",  # image level flags
             "imageHeight",
             "imageWidth",
+            "caption",
         ]
         shape_keys = [
             "label",
@@ -93,6 +95,10 @@ def load(self, filename):
                 data = json.load(f)
             imageData = None
             imagePath = None
+            caption = None
+
+            if data['caption'] is not None:
+                caption = data['caption']
 
             if data["imageData"] is not None:
                 imageData = base64.b64decode(data["imageData"])
@@ -142,6 +148,10 @@ def load(self, filename):
         self.imageData = imageData
         self.filename = filename
         self.otherData = otherData
+        self.caption = caption
+
+    def get_caption(self):
+        return self.caption
 
     @staticmethod
     def _check_image_height_and_width(imageData, imageHeight, imageWidth):
@@ -170,6 +180,7 @@ def save(
         imageData=None,
         otherData=None,
         flags=None,
+        caption=None,
     ):
         if imageData is not None:
             imageData = base64.b64encode(imageData).decode("utf-8")
@@ -188,6 +199,7 @@ def save(
             imageData=imageData,
             imageHeight=imageHeight,
             imageWidth=imageWidth,
+            caption=caption,
         )
         for key, value in otherData.items():
             assert key not in data

annolid/gui/widgets/canvas.py

@@ -54,6 +54,7 @@ def __init__(self, *args, **kwargs):
                     self.double_click
                 )
             )
+        self.caption_label = QtWidgets.QTextEdit()
         self.num_backups = kwargs.pop("num_backups", 10)
         self._crosshair = kwargs.pop(
             "crosshair",
@@ -138,6 +139,12 @@ def fillDrawing(self):
     def setFillDrawing(self, value):
         self._fill_drawing = value
 
+    def setCaption(self, text):
+        self.caption_label.setText(text)
+
+    def getCaption(self):
+        return self.caption_label.toPlainText()
+
     @property
     def createMode(self):
         return self._createMode
@@ -1013,6 +1020,34 @@ def paintEvent(self, event):
         p.drawPixmap(0, 0, self.pixmap)
         self.sam_mask.paint(p)
 
+        # Draw text from the label at the desired position
+        if self.caption_label.toPlainText():
+            text = self.caption_label.toPlainText()
+
+            # Define a bounding rectangle for the text (adjust the position and size as needed)
+            rect = QtCore.QRect(5, 3, 500, 55)  # x, y, width, height
+
+            # Set the background color (optional)
+            background_color = QtGui.QColor(
+                255, 255, 255, 128)  # White background
+            # Fill the rectangle with the background color
+            p.fillRect(rect, background_color)
+
+            # Set the font for the text
+            # Change to your preferred font and size
+            font = QtGui.QFont("Arial", 8)
+            p.setFont(font)
+
+            # Set text color
+            text_color = QtGui.QColor(0, 0, 0)  # Black text color
+            p.setPen(text_color)
+
+            # Set text alignment and enable word wrapping
+            alignment = QtCore.Qt.AlignLeft | QtCore.Qt.AlignTop  # Adjust alignment as needed
+
+            # Draw the text within the rectangle with word wrapping
+            p.drawText(rect, alignment | QtCore.Qt.TextWordWrap, text)
+
         # draw crosshair
         if ((not self.createMode == 'grounding_sam')
             and (self._crosshair[self._createMode]