BUG: infer_objects with Intervals

doc/source/whatsnew/v2.0.0.rst

@@ -704,6 +704,7 @@ Strings
 Interval
 ^^^^^^^^
 - Bug in :meth:`IntervalIndex.is_overlapping` incorrect output if interval has duplicate left boundaries (:issue:`49581`)
+- Bug in :meth:`Series.infer_objects` failing to infer :class:`IntervalDtype` for an object series of :class:`Interval` objects (:issue:`50090`)
 -
 
 Indexing

pandas/core/internals/array_manager.py

@@ -384,6 +384,7 @@ def _convert(arr):
                     convert_datetime=True,
                     convert_timedelta=True,
                     convert_period=True,
+                    convert_interval=True,
                 )
             else:
                 return arr.copy()

pandas/core/internals/blocks.py

@@ -1980,6 +1980,7 @@ def convert(
             convert_datetime=True,
             convert_timedelta=True,
             convert_period=True,
+            convert_interval=True,
         )
         res_values = ensure_block_shape(res_values, self.ndim)
         return [self.make_block(res_values)]

pandas/tests/series/methods/test_infer_objects.py

@@ -1,5 +1,6 @@
 import numpy as np
 
+from pandas import interval_range
 import pandas._testing as tm
 
 
@@ -22,3 +23,11 @@ def test_infer_objects_series(self, index_or_series):
 
         assert actual.dtype == "object"
         tm.assert_equal(actual, expected)
+
+    def test_infer_objects_interval(self, index_or_series):
+        # GH#50090
+        ii = interval_range(1, 10)
+        obj = index_or_series(ii)
+
+        result = obj.astype(object).infer_objects()
+        tm.assert_equal(result, obj)