Update filtering skill up page

CHANGELOG.md

@@ -12,7 +12,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - Develop JsonSectionPageMapper in Rust API
   (<https://github.com/openvinotoolkit/datumaro/pull/1224>)
 - Add Filtering via User-Provided Python Functions
-  (<https://github.com/openvinotoolkit/datumaro/pull/1230>)
+  (<https://github.com/openvinotoolkit/datumaro/pull/1230>, <https://github.com/openvinotoolkit/datumaro/pull/1233>)
 
 ### Enhancements
 - Optimize Python import to make CLI entrypoint faster

docs/source/docs/level-up/intermediate_skills/09_data_filtering.rst

@@ -11,9 +11,13 @@ of data continue to grow, data filtering will become an increasingly important a
 By filtering the dataset in this way, we can create a subset of data that is tailored to our specific needs, making it easier
 to extract meaningful insights or use it effectively for decision-making purposes.
 
-In this tutorial, we provide the simple example of filtering dataset using item and annotation. To set how to filter dataset,
-which satisfied some condition, we use XML as query format. Refer this `XPATH <https://devhints.io/xpath>`_ to set your own filter.
-The detailed description for filter operation is given by :doc:`Filter <../../command-reference/context_free/filter>`.
+In this tutorial, we provide the simple example of filtering dataset items and annotations.
+To set the filtering condition, we can use
+
+1) [**ProjectCLI**, **CLI**, **Python**] `XPATH <https://devhints.io/xpath>`_ query,
+2) [**Python**] User-provided Python function query.
+
+The detailed description for the `XPATH <https://devhints.io/xpath>`_ query is given by :doc:`this page <../../command-reference/context_free/filter>`.
 The more advanced Python example is given :doc:`this notebook <../../jupyter_notebook_examples/notebooks/04_filter>`.
 
 .. tab-set::
@@ -57,9 +61,56 @@ The more advanced Python example is given :doc:`this notebook <../../jupyter_not
             from datumaro.components.dataset import Dataset
 
             dataset_path = '/path/to/data'
-            dataset = Dataset.import_from(dataset_path, 'datumaro')
+            dataset = Dataset.import_from(dataset_path, format='datumaro')
 
             filtered_result = Dataset.filter(dataset, 'how/to/filter/dataset')
 
         We can set ``<how/to/filter/dataset>`` as your own filter like ``'/item/annotation[occluded="True"]'``.
         This example command will filter only items through the annotation attribute which has `occluded`.
+
+        In addition, you can filter dataset items with your custom Python fuction as well.
+        For example, an example of filtering dataset items with images larger than 1024 pixels:
+
+        .. code-block:: python
+
+            from datumaro.components.dataset import Dataset
+            from datumaro.components.media import Image
+
+            def filter_func(item: DatasetItem) -> bool:
+                h, w = item.media_as(Image).size
+                return h > 1024 or w > 1024
+
+            dataset_path = '/path/to/data'
+            dataset = Dataset.import_from(dataset_path, format='datumaro')
+
+            filtered_result = Dataset.filter(dataset, filter_func)
+
+        On the other hand, it is possible to filter dataset annotations with the user-provided Python function.
+        This is an example of removing bounding boxes sized greater than 50% of the image size:
+
+        .. code-block:: python
+
+            from datumaro.components.dataset import Dataset
+            from datumaro.components.media import Image
+            from datumaro.components.annotation import Annotation, Bbox
+
+            def filter_func(item: DatasetItem, ann: Annotation) -> bool:
+                # If the annotation is not a Bbox, do not filter
+                if not isinstance(ann, Bbox):
+                    return False
+
+                h, w = item.media_as(Image).size
+                image_size = h * w
+                bbox_size = ann.h * ann.w
+
+                # Accept Bboxes smaller than 50% of the image size
+                return bbox_size < 0.5 * image_size
+
+            def filter_func(item: DatasetItem) -> bool:
+                h, w = item.media_as(Image).size
+                return h > 1024 or w > 1024
+
+            dataset_path = '/path/to/data'
+            dataset = Dataset.import_from(dataset_path, format='datumaro')
+
+            filtered_result = Dataset.filter(dataset, filter_func)