Added docstrings for some transformers

Author: mykola

scaledp/image/ImageCropBoxes.py

@@ -59,6 +59,7 @@ class ImageCropBoxes(
         "Padding.",
         typeConverter=TypeConverters.toInt,
     )
+
     noCrop = Param(
         Params._dummy(),
         "noCrop",
@@ -73,6 +74,13 @@ class ImageCropBoxes(
         typeConverter=TypeConverters.toInt,
     )
 
+    autoRotate = Param(
+        Params._dummy(),
+        "autoRotate",
+        "Auto rotate cropped image if box height > box width.",
+        typeConverter=TypeConverters.toBoolean,
+    )
+
     defaultParams = MappingProxyType(
         {
             "inputCols": ["image", "boxes"],
@@ -85,6 +93,7 @@ class ImageCropBoxes(
             "propagateError": False,
             "noCrop": True,
             "limit": 0,
+            "autoRotate": True,
         },
     )
 
@@ -115,7 +124,7 @@ def transform_udf(self, image, data):
                 box = b
                 if not isinstance(box, Box):
                     box = Box(**box.asDict())
-                if box.width < box.height:
+                if self.getAutoRotate() and box.width < box.height:
                     cropped_image = img.crop(box.bbox(self.getPadding())).rotate(
                         -90,
                         expand=True,

scaledp/models/detectors/BaseDetector.py

@@ -75,6 +75,7 @@ def get_params(self):
         return json.dumps({k.name: v for k, v in self.extractParamMap().items()})
 
     def outputSchema(self):
+        """Output schema of the detector."""
         return StructType(
             [
                 StructField("path", StringType(), True),
@@ -92,6 +93,9 @@ def outputSchema(self):
         )
 
     def transform_udf(self, image, params=None):
+        """
+        Run detector on a single image.
+        """
         logging.info("Run Detector")
         if params is None:
             params = self.get_params()
@@ -147,6 +151,9 @@ def transform_udf_pandas(
         images: pd.DataFrame,
         params: pd.Series,
     ) -> pd.DataFrame:
+        """
+        Run detector on a batch of images.
+        """
         params = json.loads(params[0])
         resized_images = []
         for _index, img in images.iterrows():

scaledp/models/detectors/CraftTextDetector.py

@@ -12,6 +12,8 @@
 
 
 class CraftTextDetector(BaseDetector, HasDevice, HasBatchSize):
+    """CRAFT text detector."""
+
     _craft_net = None
     _refine_net = None
 

scaledp/models/detectors/YoloOnnxTextDetector.py

@@ -17,6 +17,8 @@
 
 
 class YoloOnnxTextDetector(BaseDetector, HasDevice, HasBatchSize):
+    """YOLO ONNX text detector."""
+
     _model = None
 
     task = Param(

scaledp/models/detectors/__init__.py

@@ -0,0 +1,17 @@
+from scaledp.models.detectors.BaseDetector import BaseDetector
+from scaledp.models.detectors.CraftTextDetector import CraftTextDetector
+from scaledp.models.detectors.DBNetOnnxDetector import DBNetOnnxDetector
+from scaledp.models.detectors.FaceDetector import FaceDetector
+from scaledp.models.detectors.SignatureDetector import SignatureDetector
+from scaledp.models.detectors.YoloOnnxDetector import YoloOnnxDetector
+from scaledp.models.detectors.YoloOnnxTextDetector import YoloOnnxTextDetector
+
+__all__ = [
+    "FaceDetector",
+    "SignatureDetector",
+    "YoloOnnxDetector",
+    "BaseDetector",
+    "DBNetOnnxDetector",
+    "CraftTextDetector",
+    "YoloOnnxTextDetector",
+]

scaledp/pdf/PdfDataToSingleImage.py

@@ -65,7 +65,7 @@ def process(self, input, path, resolution=0):
         except Exception:
             return Image(
                 path,
-                exception="Error in extration of image from pdf document",
+                exception="Error in extraction of image from pdf document",
             )
 
     def getPageNumber(self):