Merge pull request #36 from Giskard-AI/meta-wrapper

Refactoring with meta wrapper

Author: rabah-khalek

examples/landmark_detection/criterias/criteria3_face_orientations.ipynb

@@ -48,7 +48,7 @@
     "cached_dl = CachedDataLoader(HeadPoseDataLoader(dl), cache_size=None, cache_img=False, cache_labels=False)\n",
     "\n",
     "for idx, (img, marks, meta) in enumerate(cached_dl):\n",
-    "    print(idx, meta[0][\"headPose\"])"
+    "    print(idx, meta[0])"
    ]
   },
   {
@@ -57,7 +57,7 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "head_pose_dl = FilteredDataLoader(cached_dl, lambda elt: elt[2][\"headPose\"][\"roll\"] > 0)"
+    "head_pose_dl = FilteredDataLoader(cached_dl, lambda elt: elt[2].get_includes(\"roll\") > 0)"
    ]
   },
   {
@@ -78,7 +78,7 @@
    ],
    "source": [
     "for idx, (img, marks, meta) in enumerate(head_pose_dl):\n",
-    "    print(head_pose_dl._reindex[idx], meta[0][\"headPose\"])"
+    "    print(head_pose_dl._reindex[idx], meta[0])"
    ]
   },
   {

examples/landmark_detection/criterias/criteria4_ethnicity.ipynb

@@ -42,7 +42,7 @@
     "cached_dl = CachedDataLoader(ethnicity_dl, cache_size=None, cache_img=False, cache_labels=False)\n",
     "dl.name\n",
     "\n",
-    "asians = FilteredDataLoader(cached_dl, lambda elt: elt[2][\"ethnicity\"] == \"asian\")\n",
+    "asians = FilteredDataLoader(cached_dl, lambda elt: elt[2].get_includes(\"ethnicity\") == \"asian\")\n",
     "asians._reindex"
    ]
   }

examples/landmark_detection/demo/ethnicity_criteria.ipynb

@@ -54,8 +54,8 @@
     "cached_dl = CachedDataLoader(ethnicity_dl, cache_size=None, cache_img=False, cache_labels=False)\n",
     "dl.name\n",
     "\n",
-    "asians = FilteredDataLoader(cached_dl, lambda elt: elt[2][\"ethnicity\"] == \"asian\")\n",
-    "whites = FilteredDataLoader(cached_dl, lambda elt: elt[2][\"ethnicity\"] == \"white\")"
+    "asians = FilteredDataLoader(cached_dl, lambda elt: elt[2].get_includes(\"ethnicity\") == \"asian\")\n",
+    "whites = FilteredDataLoader(cached_dl, lambda elt: elt[2].get_includes(\"ethnicity\") == \"white\")"
    ]
   },
   {

examples/landmark_detection/demo/report_example.ipynb

@@ -238,11 +238,11 @@
     "\n",
     "\n",
     "def positive_roll(elt):\n",
-    "    return elt[2][\"headPose\"][\"roll\"] > 0\n",
+    "    return elt[2].get_includes(\"roll\") > 0\n",
     "\n",
     "\n",
     "def negative_roll(elt):\n",
-    "    return elt[2][\"headPose\"][\"roll\"] < 0\n",
+    "    return elt[2].get_includes(\"roll\") < 0\n",
     "\n",
     "\n",
     "head_poses = [positive_roll, negative_roll]\n",
@@ -288,7 +288,7 @@
     "\n",
     "\n",
     "def white_ethnicity(elt):\n",
-    "    return elt[2][\"ethnicity\"] == \"white\"\n",
+    "    return elt[2].get_includes(\"ethnicity\") == \"white\"\n",
     "\n",
     "\n",
     "ethnicities = [white_ethnicity]\n",

examples/landmark_detection/ffhq/ffhq-filtering-caching.ipynb

@@ -67,11 +67,11 @@
     "\n",
     "# head pose filtering\n",
     "def positive_roll(elt):\n",
-    "    return elt[2][\"headPose\"][\"roll\"] > 0\n",
+    "    return elt[2].get_includes(\"roll\") > 0\n",
     "\n",
     "\n",
     "def negative_roll(elt):\n",
-    "    return elt[2][\"headPose\"][\"roll\"] < 0\n",
+    "    return elt[2].get_includes(\"roll\") < 0\n",
     "\n",
     "\n",
     "cached_dl = CachedDataLoader(HeadPoseDataLoader(dl_ref), cache_size=None, cache_img=False, cache_labels=False)\n",
@@ -81,11 +81,11 @@
     "\n",
     "# ethnicity filtering\n",
     "def white_ethnicity(elt):\n",
-    "    return elt[2][\"ethnicity\"] == \"white\"\n",
+    "    return elt[2].get_includes(\"ethnicity\") == \"white\"\n",
     "\n",
     "\n",
     "def latino_ethnicity(elt):\n",
-    "    return elt[2][\"ethnicity\"] == \"latino hispanic\"\n",
+    "    return elt[2].get_includes(\"ethnicity\") == \"latino hispanic\"\n",
     "\n",
     "\n",
     "cached_dl = CachedDataLoader(\n",

examples/landmark_detection/ffhq/ffhq-filtering.ipynb

@@ -1,11 +1,10 @@
 import math
 from abc import ABC, abstractmethod
-from typing import Any, Dict, List, Optional, Tuple
+from typing import List, Optional
 
 import numpy as np
 
-SingleLabel = Tuple[np.ndarray, Any, Optional[Dict[Any, Any]]]
-BatchedLabels = Tuple[Tuple[np.ndarray], Any, Tuple[Optional[Dict[Any, Any]]]]
+from ..types import TypesBase
 
 
 class DataIteratorBase(ABC):
@@ -104,7 +103,7 @@ def labels_none(self) -> Optional[np.ndarray]:
         """
         return None
 
-    def meta_none(self) -> Optional[Dict]:
+    def meta_none(self) -> Optional[TypesBase.meta]:
         """
         Returns default for meta data if it is None.
 
@@ -125,15 +124,15 @@ def get_labels(self, idx: int) -> Optional[np.ndarray]:
         """
         return None
 
-    def get_meta(self, idx: int) -> Optional[Dict]:
+    def get_meta(self, idx: int) -> Optional[TypesBase.meta]:
         """
         Gets meta information (for a single image) for a specific index.
 
         Args:
             idx (int): Index of the image.
 
         Returns:
-            Optional[Dict]: Meta information for the given index.
+            Optional[TypesBase.meta]: Meta information for the given index.
         """
         return None
 
@@ -163,27 +162,27 @@ def get_meta_with_default(self, idx: int) -> np.ndarray:
         meta = self.get_meta(idx)
         return meta if meta is not None else self.meta_none()
 
-    def get_single_element(self, idx) -> SingleLabel:
+    def get_single_element(self, idx) -> TypesBase.single_data:
         """
         Gets a single element as a tuple of (image, labels, meta) for a specific index.
 
         Args:
             idx (int): Index of the image.
 
         Returns:
-            SingleLabel: Tuple containing image, labels, and meta information.
+            TypesBase.single_data: Tuple containing image, labels, and meta information.
         """
         return self.get_image(idx), self.get_labels_with_default(idx), self.get_meta_with_default(idx)
 
-    def __getitem__(self, idx: int) -> BatchedLabels:
+    def __getitem__(self, idx: int) -> TypesBase.batched_data:
         """
         Gets a batch of elements for a specific index.
 
         Args:
             idx (int): Index of the batch.
 
         Returns:
-            BatchedLabels: Batched data containing images, labels, and meta information.
+            TypesBase.batched_data: Batched data containing images, labels, and meta information.
         """
         return self._collate_fn(
             [self.get_single_element(i) for i in self.idx_sampler[idx * self.batch_size : (idx + 1) * self.batch_size]]
@@ -220,28 +219,28 @@ def all_meta(self) -> List:
         """
         return [self.get_meta_with_default(idx) for idx in self.idx_sampler]
 
-    def __next__(self) -> BatchedLabels:
+    def __next__(self) -> TypesBase.batched_data:
         """
         Gets the next batch of elements.
 
         Returns:
-            BatchedLabels: Batched data containing images, labels, and meta information.
+            TypesBase.batched_data: Batched data containing images, labels, and meta information.
         """
         if self.idx >= len(self):
             raise StopIteration
         elt = self[self.idx]
         self.idx += 1
         return elt
 
-    def _collate_fn(self, elements: List[SingleLabel]) -> BatchedLabels:
+    def _collate_fn(self, elements: List[TypesBase.single_data]) -> TypesBase.batched_data:
         """
         Collates a list of single elements into a batched element.
 
         Args:
-            elements (List[SingleLabel]): List of single elements.
+            elements (List[TypesBase.single_data]): List of single elements.
 
         Returns:
-            BatchedLabels: Batched data containing images, labels, and meta information.
+            TypesBase.batched_data: Batched data containing images, labels, and meta information.
         """
         batched_elements = list(zip(*elements, strict=True))
         return batched_elements[0], np.array(batched_elements[1]), batched_elements[2]
@@ -311,15 +310,15 @@ def get_labels(self, idx: int) -> Optional[np.ndarray]:
         """
         return self._wrapped_dataloader.get_labels(idx)
 
-    def get_meta(self, idx: int) -> Optional[Dict]:
+    def get_meta(self, idx: int) -> Optional[TypesBase.meta]:
         """
         Gets meta information from the wrapped data loader.
 
         Args:
             idx (int): Index of the data.
 
         Returns:
-            Optional[Dict]: Meta information from the wrapped data loader.
+            Optional[TypesBase.meta]: Meta information from the wrapped data loader.
         """
         return self._wrapped_dataloader.get_meta(idx)