Add support for selecting/excluding group slices

docs/source/user_guide/groups.rst

@@ -815,8 +815,8 @@ Selecting slices
 
 You can use
 :meth:`select_group_slices() <fiftyone.core.collections.SampleCollection.select_group_slices>`
-to create *non-grouped views* that contain one or more slices of data from a
-grouped dataset.
+to select one or more slices of data from a grouped dataset, either as a
+grouped view or as a flattened *non-grouped* view.
 
 For example, you can create an image view that contains only the left camera
 images from the grouped dataset:
@@ -843,7 +843,7 @@ images from the grouped dataset:
     View stages:
         1. SelectGroupSlices(slices='left')
 
-or you could create an image collection containing the left and right camera
+or you can create an image collection containing the left and right camera
 images:
 
 .. code-block:: python
@@ -882,30 +882,59 @@ the fact that their data is sourced from a grouped dataset!
 
     # Add fields/tags, run evaluation, export, etc
 
-Also note that any filtering that you apply prior to a
+.. note::
+
+    Any filtering that you apply prior to a
+    :meth:`select_group_slices() <fiftyone.core.collections.SampleCollection.select_group_slices>`
+    stage in a view is **not** automatically reflected by the output view, as
+    the stage looks up unfiltered slice data from the source collection:
+
+    .. code-block:: python
+
+        # Filter the active slice to locate groups of interest
+        match_view = dataset.filter_labels(...).match(...)
+
+        # Lookup all image slices for the matching groups
+        # This view contains *unfiltered* image slices
+        images_view = match_view.select_group_slices(media_type="image")
+
+    Instead, you can apply the same (or different) filtering *after* the
+    :meth:`select_group_slices() <fiftyone.core.collections.SampleCollection.select_group_slices>`
+    stage:
+
+    .. code-block:: python
+
+        # Now apply filters to the flattened collection
+        match_images_view = images_view.filter_labels(...).match(...)
+
+Alternatively, you can pass `flat=False` to
 :meth:`select_group_slices() <fiftyone.core.collections.SampleCollection.select_group_slices>`
-stage in a view is **not** automatically reflected by the output view, as the
-stage looks up unfiltered slice data from the source collection:
+to create a grouped view that only contains certain group slices:
 
 .. code-block:: python
     :linenos:
 
-    # Filter the active slice to locate groups of interest
-    match_view = dataset.filter_labels(...).match(...)
+    no_center_view = dataset.select_group_slices(["left", "right"], flat=False)
 
-    # Lookup all image slices for the matching groups
-    # This view contains *unfiltered* image slices
-    images_view = match_view.select_group_slices(media_type="image")
+    assert no_center_view.media_type == "group"
+    assert no_center_view.group_slices == ["left", "right"]
 
-Instead, you can apply the same (or different) filtering *after* the
-:meth:`select_group_slices() <fiftyone.core.collections.SampleCollection.select_group_slices>`
-stage:
+.. _groups-excluding-slices:
+
+Excluding slices
+----------------
+
+You can use
+:meth:`exclude_group_slices() <fiftyone.core.collections.SampleCollection.exclude_group_slices>`
+to create a grouped view that excludes certain slice(s) of a grouped dataset:
 
 .. code-block:: python
     :linenos:
 
-    # Now apply filters to the flattened collection
-    match_images_view = images_view.filter_labels(...).match(...)
+    no_center_view = dataset.exclude_group_slices("center")
+
+    assert no_center_view.media_type == "group"
+    assert no_center_view.group_slices == ["left", "right"]
 
 .. _groups-aggregations:
 

fiftyone/__public__.py

@@ -198,6 +198,7 @@
     ExcludeFields,
     ExcludeFrames,
     ExcludeGroups,
+    ExcludeGroupSlices,
     ExcludeLabels,
     Exists,
     FilterField,

fiftyone/core/collections.py

@@ -4618,6 +4618,80 @@ def exclude_groups(self, group_ids):
         """
         return self._add_view_stage(fos.ExcludeGroups(group_ids))
 
+    @view_stage
+    def exclude_group_slices(self, slices=None, media_type=None):
+        """Excludes the specified group slice(s) from the grouped collection.
+
+        Examples::
+
+            import fiftyone as fo
+
+            dataset = fo.Dataset()
+            dataset.add_group_field("group", default="ego")
+
+            group1 = fo.Group()
+            group2 = fo.Group()
+
+            dataset.add_samples(
+                [
+                    fo.Sample(
+                        filepath="/path/to/left-image1.jpg",
+                        group=group1.element("left"),
+                    ),
+                    fo.Sample(
+                        filepath="/path/to/video1.mp4",
+                        group=group1.element("ego"),
+                    ),
+                    fo.Sample(
+                        filepath="/path/to/right-image1.jpg",
+                        group=group1.element("right"),
+                    ),
+                    fo.Sample(
+                        filepath="/path/to/left-image2.jpg",
+                        group=group2.element("left"),
+                    ),
+                    fo.Sample(
+                        filepath="/path/to/video2.mp4",
+                        group=group2.element("ego"),
+                    ),
+                    fo.Sample(
+                        filepath="/path/to/right-image2.jpg",
+                        group=group2.element("right"),
+                    ),
+                ]
+            )
+
+            #
+            # Exclude the samples from the "ego" group slice
+            #
+
+            view = dataset.exclude_group_slices("ego")
+
+            #
+            # Exclude the samples from the "left" or "right" group slices
+            #
+
+            view = dataset.exclude_group_slices(["left", "right"])
+
+            #
+            # Exclude all image slices
+            #
+
+            view = dataset.exclude_group_slices(media_type="image")
+
+        Args:
+            slices (None): a group slice or iterable of group slices to
+                exclude
+            media_type (None): a media type or iterable of media types whose
+                slice(s) to exclude
+
+        Returns:
+            a :class:`fiftyone.core.view.DatasetView`
+        """
+        return self._add_view_stage(
+            fos.ExcludeGroupSlices(slices=slices, media_type=media_type)
+        )
+
     @view_stage
     def exclude_labels(
         self, labels=None, ids=None, tags=None, fields=None, omit_empty=True
@@ -6594,19 +6668,24 @@ def select_group_slices(
         self,
         slices=None,
         media_type=None,
+        flat=True,
         _allow_mixed=False,
         _force_mixed=False,
     ):
-        """Selects the samples in the group collection from the given slice(s).
+        """Selects the specified group slice(s) from the grouped collection.
+
+        When ``flat==True``, the returned view is a flattened non-grouped view
+        containing the samples from the slice(s) of interest.
 
-        The returned view is a flattened non-grouped view containing only the
-        slice(s) of interest.
+        When ``flat=False``, the returned view is a grouped collection
+        containing only the slice(s) of interest.
 
         .. note::
 
-            This stage performs a ``$lookup`` that pulls the requested slice(s)
-            for each sample in the input collection from the source dataset.
-            As a result, this stage always emits *unfiltered samples*.
+            When ``flat=True``, this stage performs a ``$lookup`` that pulls
+            the requested slice(s) for each sample in the input collection from
+            the source dataset. As a result, the stage emits
+            *unfiltered samples*.
 
         Examples::
 
@@ -6659,6 +6738,12 @@ def select_group_slices(
 
             view = dataset.select_group_slices(["left", "right"])
 
+            #
+            # Select only the "left" and "right" group slices
+            #
+
+            view = dataset.select_group_slices(["left", "right"], flat=False)
+
             #
             # Retrieve all image samples
             #
@@ -6669,7 +6754,10 @@ def select_group_slices(
             slices (None): a group slice or iterable of group slices to select.
                 If neither argument is provided, a flattened list of all
                 samples is returned
-            media_type (None): a media type whose slice(s) to select
+            media_type (None): a media type or iterable of media types whose
+                slice(s) to select
+            flat (True): whether to return a flattened collection (True) or a
+                grouped collection (False)
 
         Returns:
             a :class:`fiftyone.core.view.DatasetView`
@@ -6678,6 +6766,7 @@ def select_group_slices(
             fos.SelectGroupSlices(
                 slices=slices,
                 media_type=media_type,
+                flat=flat,
                 _allow_mixed=_allow_mixed,
                 _force_mixed=_force_mixed,
             )
@@ -10542,24 +10631,23 @@ def _contains_media_type(self, media_type, any_slice=False):
             return True
 
         if self.media_type == fom.GROUP:
-            if self.group_media_types is None:
+            group_media_types = self.group_media_types
+            if group_media_types is None:
                 return self._dataset.media_type == media_type
 
             if any_slice:
                 return any(
                     slice_media_type == media_type
-                    for slice_media_type in self.group_media_types.values()
+                    for slice_media_type in group_media_types.values()
                 )
 
-            return (
-                self.group_media_types.get(self.group_slice, None)
-                == media_type
-            )
+            return group_media_types.get(self.group_slice, None) == media_type
 
         if self.media_type == fom.MIXED:
+            group_media_types = self._get_group_media_types()
             return any(
                 slice_media_type == media_type
-                for slice_media_type in self._get_group_media_types().values()
+                for slice_media_type in group_media_types.values()
             )
 
         return False

fiftyone/core/dataset.py

@@ -8484,14 +8484,16 @@ def _clone_collection(sample_collection, name, persistent):
     slug = _validate_dataset_name(name)
 
     contains_videos = sample_collection._contains_videos(any_slice=True)
-    contains_groups = sample_collection.media_type == fom.GROUP
 
     if isinstance(sample_collection, fov.DatasetView):
         dataset = sample_collection._dataset
         view = sample_collection
 
         if view.media_type == fom.MIXED:
             raise ValueError("Cloning mixed views is not allowed")
+
+        if view._is_dynamic_groups:
+            raise ValueError("Cloning dynamic grouped views is not allowed")
     else:
         dataset = sample_collection
         view = None
@@ -8525,10 +8527,9 @@ def _clone_collection(sample_collection, name, persistent):
     dataset_doc.sample_collection_name = sample_collection_name
     dataset_doc.frame_collection_name = frame_collection_name
     dataset_doc.media_type = sample_collection.media_type
-    if not contains_groups:
-        dataset_doc.group_field = None
-        dataset_doc.group_media_types = {}
-        dataset_doc.default_group_slice = None
+    dataset_doc.group_field = sample_collection.group_field
+    dataset_doc.group_media_types = sample_collection.group_media_types
+    dataset_doc.default_group_slice = sample_collection.default_group_slice
 
     for field in dataset_doc.sample_fields:
         field._set_created_at(now)