clarifai 11.3.0rc2__py3-none-any.whl → 11.4.0__py3-none-any.whl

clarifai/datasets/upload/loaders/coco_detection.py

@@ -3,78 +3,85 @@
 import os
 
 from ..base import ClarifaiDataLoader
-
 from ..features import VisualDetectionFeatures
 
-#pycocotools is a dependency for this loader
+# pycocotools is a dependency for this loader
 try:
-  from pycocotools.coco import COCO
+    from pycocotools.coco import COCO
 except ImportError:
-  raise ImportError("Could not import pycocotools package. "
-                    "Please do `pip install 'clarifai[all]'` to import pycocotools.")
+    raise ImportError(
+        "Could not import pycocotools package. "
+        "Please do `pip install 'clarifai[all]'` to import pycocotools."
+    )
 
 
 class COCODetectionDataLoader(ClarifaiDataLoader):
-
-  def __init__(self, images_dir, label_filepath):
-    """
-      Args:
-        images_dir: Directory containing the images.
-        label_filepath: Path to the COCO annotation file.
-      """
-    self.images_dir = images_dir
-    self.label_filepath = label_filepath
-
-    self.map_ids = {}
-    self.load_data()
-
-  @property
-  def task(self):
-    return "visual_detection"
-
-  def load_data(self) -> None:
-    self.coco = COCO(self.label_filepath)
-    self.map_ids = {i: img_id for i, img_id in enumerate(list(self.coco.imgs.keys()))}
-
-  def __getitem__(self, index: int):
-    value = self.coco.imgs[self.map_ids[index]]
-    image_path = os.path.join(self.images_dir, value['file_name'])
-    annots = []  # bboxes
-    concept_ids = []
-
-    input_ann_ids = self.coco.getAnnIds(imgIds=[value['id']])
-    input_anns = self.coco.loadAnns(input_ann_ids)
-
-    for ann in input_anns:
-      # get concept info
-      # note1: concept_name can be human readable
-      # note2: concept_id can only be alphanumeric, up to 32 characters, with no special chars except `-` and `_`
-      concept_name = self.coco.cats[ann['category_id']]['name']
-      concept_id = concept_name.lower().replace(' ', '-')
-
-      # get bbox information
-      # note1: coco bboxes are `[x_min, y_min, width, height]` in pixels
-      # note2: clarifai bboxes are `[x_min, y_min, x_max, y_max]` normalized between 0-1.0
-      coco_bbox = ann['bbox']
-      clarifai_bbox = {
-          'left_col': max(0, coco_bbox[0] / value['width']),
-          'top_row': max(0, coco_bbox[1] / value['height']),
-          'right_col': min(1, (coco_bbox[0] + coco_bbox[2]) / value['width']),
-          'bottom_row': min(1, (coco_bbox[1] + coco_bbox[3]) / value['height'])
-      }
-      if (clarifai_bbox['left_col'] >=
-          clarifai_bbox['right_col']) or (clarifai_bbox['top_row'] >= clarifai_bbox['bottom_row']):
-        continue
-      annots.append([
-          clarifai_bbox['left_col'], clarifai_bbox['top_row'], clarifai_bbox['right_col'],
-          clarifai_bbox['bottom_row']
-      ])
-      concept_ids.append(concept_id)
-
-    assert len(concept_ids) == len(annots), f"Num concepts must match num bbox annotations\
+    def __init__(self, images_dir, label_filepath):
+        """
+        Args:
+          images_dir: Directory containing the images.
+          label_filepath: Path to the COCO annotation file.
+        """
+        self.images_dir = images_dir
+        self.label_filepath = label_filepath
+
+        self.map_ids = {}
+        self.load_data()
+
+    @property
+    def task(self):
+        return "visual_detection"
+
+    def load_data(self) -> None:
+        self.coco = COCO(self.label_filepath)
+        self.map_ids = {i: img_id for i, img_id in enumerate(list(self.coco.imgs.keys()))}
+
+    def __getitem__(self, index: int):
+        value = self.coco.imgs[self.map_ids[index]]
+        image_path = os.path.join(self.images_dir, value['file_name'])
+        annots = []  # bboxes
+        concept_ids = []
+
+        input_ann_ids = self.coco.getAnnIds(imgIds=[value['id']])
+        input_anns = self.coco.loadAnns(input_ann_ids)
+
+        for ann in input_anns:
+            # get concept info
+            # note1: concept_name can be human readable
+            # note2: concept_id can only be alphanumeric, up to 32 characters, with no special chars except `-` and `_`
+            concept_name = self.coco.cats[ann['category_id']]['name']
+            concept_id = concept_name.lower().replace(' ', '-')
+
+            # get bbox information
+            # note1: coco bboxes are `[x_min, y_min, width, height]` in pixels
+            # note2: clarifai bboxes are `[x_min, y_min, x_max, y_max]` normalized between 0-1.0
+            coco_bbox = ann['bbox']
+            clarifai_bbox = {
+                'left_col': max(0, coco_bbox[0] / value['width']),
+                'top_row': max(0, coco_bbox[1] / value['height']),
+                'right_col': min(1, (coco_bbox[0] + coco_bbox[2]) / value['width']),
+                'bottom_row': min(1, (coco_bbox[1] + coco_bbox[3]) / value['height']),
+            }
+            if (clarifai_bbox['left_col'] >= clarifai_bbox['right_col']) or (
+                clarifai_bbox['top_row'] >= clarifai_bbox['bottom_row']
+            ):
+                continue
+            annots.append(
+                [
+                    clarifai_bbox['left_col'],
+                    clarifai_bbox['top_row'],
+                    clarifai_bbox['right_col'],
+                    clarifai_bbox['bottom_row'],
+                ]
+            )
+            concept_ids.append(concept_id)
+
+        assert len(concept_ids) == len(annots), (
+            f"Num concepts must match num bbox annotations\
         for a single image. Found {len(concept_ids)} concepts and {len(annots)} bboxes."
+        )
 
-    return VisualDetectionFeatures(image_path, concept_ids, annots, id=str(value['id']))
+        return VisualDetectionFeatures(image_path, concept_ids, annots, id=str(value['id']))
 
-  def __len__(self):
-    return len(self.coco.imgs)
+    def __len__(self):
+        return len(self.coco.imgs)

clarifai/datasets/upload/loaders/imagenet_classification.py

@@ -3,60 +3,64 @@
 import os
 
 from clarifai.datasets.upload.base import ClarifaiDataLoader
+
 from ..features import VisualClassificationFeatures
 
 
 class ImageNetDataLoader(ClarifaiDataLoader):
-  """ImageNet Dataset."""
-
-  def __init__(self, data_dir, split: str = "train"):
-    """
-    Initialize dataset params.
-    Args:
-      data_dir: the local dataset directory.
-      split: "train" or "test"
-    """
-    self.split = split
-    self.data_dir = data_dir
-    self.label_map = dict()
-    self.concepts = []
-    self.image_paths = []
-
-    self.load_data()
-
-  @property
-  def task(self):
-    return "visual_classification"
-
-  def load_data(self):
-    #Creating label map
-    with open(os.path.join(self.data_dir, "LOC_synset_mapping.txt")) as _file:
-      for _id in _file:
-        #Removing the spaces,upper quotes and Converting to set to remove repetitions. Then converting to list for compatibility.
-        self.label_map[_id.split(" ")[0]] = list({
-            "".join(("".join((label.rstrip().lstrip().split(" ")))).split("'"))
-            for label in _id[_id.find(" ") + 1:].split(",")
-        })
-
-    for _folder in os.listdir(os.path.join(self.data_dir, self.split)):
-      try:
-        concept = self.label_map[_folder]  #concepts
-      except Exception:
-        continue
-      folder_path = os.path.join(self.data_dir, self.split) + "/" + _folder
-      for _img in os.listdir(folder_path):
-        if _img.lower().endswith(('.png', '.jpg', '.jpeg', '.tiff')):
-          self.concepts.append(concept)
-          self.image_paths.append(folder_path + "/" + _img)
-
-    assert len(self.concepts) == len(self.image_paths)
-    "Number of concepts and images are not equal"
-
-  def __len__(self):
-    return len(self.image_paths)
-
-  def __getitem__(self, idx):
-    return VisualClassificationFeatures(
-        image_path=self.image_paths[idx],
-        labels=self.concepts[idx],
-        id=self.image_paths[idx].split('.')[0].split('/')[-1])
+    """ImageNet Dataset."""
+
+    def __init__(self, data_dir, split: str = "train"):
+        """
+        Initialize dataset params.
+        Args:
+          data_dir: the local dataset directory.
+          split: "train" or "test"
+        """
+        self.split = split
+        self.data_dir = data_dir
+        self.label_map = dict()
+        self.concepts = []
+        self.image_paths = []
+
+        self.load_data()
+
+    @property
+    def task(self):
+        return "visual_classification"
+
+    def load_data(self):
+        # Creating label map
+        with open(os.path.join(self.data_dir, "LOC_synset_mapping.txt")) as _file:
+            for _id in _file:
+                # Removing the spaces,upper quotes and Converting to set to remove repetitions. Then converting to list for compatibility.
+                self.label_map[_id.split(" ")[0]] = list(
+                    {
+                        "".join(("".join((label.rstrip().lstrip().split(" ")))).split("'"))
+                        for label in _id[_id.find(" ") + 1 :].split(",")
+                    }
+                )
+
+        for _folder in os.listdir(os.path.join(self.data_dir, self.split)):
+            try:
+                concept = self.label_map[_folder]  # concepts
+            except Exception:
+                continue
+            folder_path = os.path.join(self.data_dir, self.split) + "/" + _folder
+            for _img in os.listdir(folder_path):
+                if _img.lower().endswith(('.png', '.jpg', '.jpeg', '.tiff')):
+                    self.concepts.append(concept)
+                    self.image_paths.append(folder_path + "/" + _img)
+
+        assert len(self.concepts) == len(self.image_paths)
+        "Number of concepts and images are not equal"
+
+    def __len__(self):
+        return len(self.image_paths)
+
+    def __getitem__(self, idx):
+        return VisualClassificationFeatures(
+            image_path=self.image_paths[idx],
+            labels=self.concepts[idx],
+            id=self.image_paths[idx].split('.')[0].split('/')[-1],
+        )