BUG: convert_dtypes incorrectly converts byte strings to strings in 1.3+

doc/source/whatsnew/v1.3.4.rst

@@ -14,6 +14,8 @@ including other versions of pandas.
 
 Fixed regressions
 ~~~~~~~~~~~~~~~~~
+
+- Fixed regression in :meth:`DataFrame.convert_dtypes` incorrectly converts byte strings to strings (:issue:`43183`)
 - Fixed regression in :meth:`.GroupBy.agg` where it was failing silently with mixed data types along ``axis=1`` and :class:`MultiIndex` (:issue:`43209`)
 - Fixed regression in :meth:`merge` with integer and ``NaN`` keys failing with ``outer`` merge (:issue:`43550`)
 - Fixed regression in :meth:`DataFrame.corr` raising ``ValueError`` with ``method="spearman"`` on 32-bit platforms (:issue:`43588`)

pandas/core/dtypes/cast.py

@@ -1426,6 +1426,8 @@ def convert_dtypes(
         if is_string_dtype(inferred_dtype):
             if not convert_string:
                 return input_array.dtype
+            elif inferred_dtype == "bytes":
+                return pandas_dtype("object")
             else:
                 return pandas_dtype("string")
 

pandas/tests/series/methods/test_convert_dtypes.py

@@ -226,3 +226,12 @@ def test_convert_bool_dtype(self):
         # GH32287
         df = pd.DataFrame({"A": pd.array([True])})
         tm.assert_frame_equal(df, df.convert_dtypes())
+
+    def test_convert_byte_string_dtype(self):
+        # GH-43183
+        byte_str = b"binary-string"
+
+        df = pd.DataFrame(data={"A": byte_str}, index=[0])
+        result = df.convert_dtypes()
+        expected = df
+        tm.assert_frame_equal(result, expected)