[DEVX-419]: Adding ID support for labels (#338)

* ID_support_for_labels

* clarifai_extra_requires

clarifai/client/input.py

@@ -72,6 +72,7 @@ def _get_proto(input_id: str,
                  text_pb: Text = None,
                  geo_info: List = None,
                  labels: List = None,
+                 label_ids: List = None,
                  metadata: Struct = None) -> Input:
     """Create input proto for image data type.
         Args:
@@ -82,22 +83,35 @@ def _get_proto(input_id: str,
             audio_pb (Audio): The audio proto to be used for the input.
             text_pb (Text): The text proto to be used for the input.
             geo_info (list): A list of longitude and latitude for the geo point.
-            labels (list): A list of labels for the input.
+            labels (list): A list of label names for the input.
+            label_ids (list): A list of label ids for the input.
             metadata (Struct): A Struct of metadata for the input.
         Returns:
             Input: An Input object for the specified input ID.
         """
     assert geo_info is None or isinstance(
         geo_info, list), "geo_info must be a list of longitude and latitude"
     assert labels is None or isinstance(labels, list), "labels must be a list of strings"
+    assert label_ids is None or isinstance(label_ids, list), "label_ids must be a list of strings"
     assert metadata is None or isinstance(metadata, Struct), "metadata must be a Struct"
     geo_pb = resources_pb2.Geo(geo_point=resources_pb2.GeoPoint(
         longitude=geo_info[0], latitude=geo_info[1])) if geo_info else None
-    concepts=[
+    if labels:
+      if not label_ids:
+        concepts=[
             resources_pb2.Concept(
             id=f"id-{''.join(_label.split(' '))}", name=_label, value=1.)\
             for _label in labels
-        ]if labels else None
+        ]
+      else:
+        assert len(labels) == len(label_ids), "labels and label_ids must be of the same length"
+        concepts=[
+            resources_pb2.Concept(
+            id=label_id, name=_label, value=1.)\
+            for label_id, _label in zip(label_ids, labels)
+        ]
+    else:
+      concepts = None
 
     if dataset_id:
       return resources_pb2.Input(
@@ -467,13 +481,14 @@ def get_text_inputs_from_folder(folder_path: str, dataset_id: str = None,
     return input_protos
 
   @staticmethod
-  def get_bbox_proto(input_id: str, label: str, bbox: List) -> Annotation:
+  def get_bbox_proto(input_id: str, label: str, bbox: List, label_id: str = None) -> Annotation:
     """Create an annotation proto for each bounding box, label input pair.
 
     Args:
         input_id (str): The input ID for the annotation to create.
-        label (str): annotation label
+        label (str): annotation label name
         bbox (List): a list of a single bbox's coordinates. # bbox ordering: [xmin, ymin, xmax, ymax]
+        label_id (str): annotation label ID
 
     Returns:
         An annotation object for the specified input ID.
@@ -500,19 +515,22 @@ def get_bbox_proto(input_id: str, label: str, bbox: List) -> Annotation:
                 data=resources_pb2.Data(concepts=[
                     resources_pb2.Concept(
                         id=f"id-{''.join(label.split(' '))}", name=label, value=1.)
+                    if not label_id else resources_pb2.Concept(id=label_id, name=label, value=1.)
                 ]))
         ]))
 
     return input_annot_proto
 
   @staticmethod
-  def get_mask_proto(input_id: str, label: str, polygons: List[List[float]]) -> Annotation:
+  def get_mask_proto(input_id: str, label: str, polygons: List[List[float]],
+                     label_id: str = None) -> Annotation:
     """Create an annotation proto for each polygon box, label input pair.
 
     Args:
         input_id (str): The input ID for the annotation to create.
-        label (str): annotation label
+        label (str): annotation label name
         polygons (List): Polygon x,y points iterable
+        label_id (str): annotation label ID
 
     Returns:
         An annotation object for the specified input ID.
@@ -537,6 +555,7 @@ def get_mask_proto(input_id: str, label: str, polygons: List[List[float]]) -> An
                 data=resources_pb2.Data(concepts=[
                     resources_pb2.Concept(
                         id=f"id-{''.join(label.split(' '))}", name=label, value=1.)
+                    if not label_id else resources_pb2.Concept(id=label_id, name=label, value=1.)
                 ]))
         ]))
 

clarifai/datasets/upload/features.py

@@ -10,6 +10,7 @@ class TextFeatures:
   labels: List[Union[str, int]]  # List[str or int] to cater for multi-class tasks
   id: Optional[int] = None  # text_id
   metadata: Optional[dict] = None
+  label_ids: Optional[List[str]] = None
 
 
 @dataclass
@@ -21,6 +22,7 @@ class VisualClassificationFeatures:
   id: Optional[int] = None  # image_id
   metadata: Optional[dict] = None
   image_bytes: Optional[bytes] = None
+  label_ids: Optional[List[str]] = None
 
 
 @dataclass
@@ -33,6 +35,7 @@ class VisualDetectionFeatures:
   id: Optional[int] = None  # image_id
   metadata: Optional[dict] = None
   image_bytes: Optional[bytes] = None
+  label_ids: Optional[List[str]] = None
 
 
 @dataclass
@@ -45,3 +48,4 @@ class VisualSegmentationFeatures:
   id: Optional[int] = None  # image_id
   metadata: Optional[dict] = None
   image_bytes: Optional[bytes] = None
+  label_ids: Optional[List[str]] = None

clarifai/datasets/upload/image.py

@@ -32,6 +32,7 @@ def process_data_item(id):
       image_path = data_item.image_path
       labels = data_item.labels if isinstance(data_item.labels,
                                               list) else [data_item.labels]  # clarifai concept
+      label_ids = data_item.label_ids
       input_id = f"{self.dataset_id}-{uuid.uuid4().hex[:8]}" if data_item.id is None else f"{self.dataset_id}-{str(data_item.id)}"
       geo_info = data_item.geo_info
       if data_item.metadata is not None:
@@ -49,6 +50,7 @@ def process_data_item(id):
                 image_bytes=data_item.image_bytes,
                 dataset_id=self.dataset_id,
                 labels=labels,
+                label_ids=label_ids,
                 geo_info=geo_info,
                 metadata=metadata))
       else:
@@ -58,6 +60,7 @@ def process_data_item(id):
                 image_file=image_path,
                 dataset_id=self.dataset_id,
                 labels=labels,
+                label_ids=label_ids,
                 geo_info=geo_info,
                 metadata=metadata))
 
@@ -91,6 +94,12 @@ def process_data_item(id):
       metadata = Struct()
       image = data_item.image_path
       labels = data_item.labels  # list:[l1,...,ln]
+      if data_item.label_ids is not None:
+        assert len(labels) == len(
+            data_item.label_ids), "Length of labels and label_ids must be equal"
+        label_ids = data_item.label_ids
+      else:
+        label_ids = None
       bboxes = data_item.bboxes  # [[xmin,ymin,xmax,ymax],...,[xmin,ymin,xmax,ymax]]
       input_id = f"{self.dataset_id}-{uuid.uuid4().hex[:8]}" if data_item.id is None else f"{self.dataset_id}-{str(data_item.id)}"
       if data_item.metadata is not None:
@@ -120,7 +129,11 @@ def process_data_item(id):
       # one id could have more than one bbox and label
       for i in range(len(bboxes)):
         annotation_protos.append(
-            Inputs.get_bbox_proto(input_id=input_id, label=labels[i], bbox=bboxes[i]))
+            Inputs.get_bbox_proto(
+                input_id=input_id,
+                label=labels[i],
+                bbox=bboxes[i],
+                label_id=label_ids[i] if label_ids else None))
 
     with ThreadPoolExecutor(max_workers=4) as executor:
       futures = [executor.submit(process_data_item, id) for id in batch_input_ids]
@@ -152,6 +165,12 @@ def process_data_item(id):
       metadata = Struct()
       image = data_item.image_path
       labels = data_item.labels
+      if data_item.label_ids is not None:
+        assert len(labels) == len(
+            data_item.label_ids), "Length of labels and label_ids must be equal"
+        label_ids = data_item.label_ids
+      else:
+        label_ids = None
       _polygons = data_item.polygons  # list of polygons: [[[x,y],...,[x,y]],...]
       input_id = f"{self.dataset_id}-{uuid.uuid4().hex[:8]}" if data_item.id is None else f"{self.dataset_id}-{str(data_item.id)}"
       if data_item.metadata is not None:
@@ -183,7 +202,11 @@ def process_data_item(id):
       for i, _polygon in enumerate(_polygons):
         try:
           annotation_protos.append(
-              Inputs.get_mask_proto(input_id=input_id, label=labels[i], polygons=_polygon))
+              Inputs.get_mask_proto(
+                  input_id=input_id,
+                  label=labels[i],
+                  polygons=_polygon,
+                  label_id=label_ids[i] if label_ids else None))
         except IndexError:
           continue
 

clarifai/datasets/upload/loaders/coco_captions.py

@@ -2,12 +2,17 @@
 
 import os
 
-from pycocotools.coco import COCO
-
 from clarifai.datasets.upload.base import ClarifaiDataLoader
 
 from ..features import VisualClassificationFeatures
 
+#pycocotools is a dependency for this loader
+try:
+  from pycocotools.coco import COCO
+except ImportError:
+  raise ImportError("Could not import pycocotools package. "
+                    "Please do `pip install 'clarifai[all]'` to import pycocotools.")
+
 
 class COCOCaptionsDataLoader(ClarifaiDataLoader):
   """COCO Image Captioning Dataset."""

clarifai/datasets/upload/loaders/coco_detection.py

@@ -2,12 +2,17 @@
 
 import os
 
-from pycocotools.coco import COCO
-
 from ..base import ClarifaiDataLoader
 
 from ..features import VisualDetectionFeatures
 
+#pycocotools is a dependency for this loader
+try:
+  from pycocotools.coco import COCO
+except ImportError:
+  raise ImportError("Could not import pycocotools package. "
+                    "Please do `pip install 'clarifai[all]'` to import pycocotools.")
+
 
 class COCODetectionDataLoader(ClarifaiDataLoader):
 

clarifai/datasets/upload/text.py

@@ -32,6 +32,7 @@ def process_data_item(id):
       text = data_item.text
       labels = data_item.labels if isinstance(data_item.labels,
                                               list) else [data_item.labels]  # clarifai concept
+      label_ids = data_item.label_ids
       input_id = f"{self.dataset_id}-{id}" if data_item.id is None else f"{self.dataset_id}-{str(data_item.id)}"
       if data_item.metadata is not None:
         metadata.update(data_item.metadata)
@@ -43,6 +44,7 @@ def process_data_item(id):
               raw_text=text,
               dataset_id=self.dataset_id,
               labels=labels,
+              label_ids=label_ids,
               metadata=metadata))
 
     with ThreadPoolExecutor(max_workers=4) as executor: