fix duplicate categorical features added from image exif metadata in …

…RAI Vision Dashboard
microsoft · Jan 4, 2024 · 470d176 · 470d176
1 parent 2556253
commit 470d176
Show file tree

Hide file tree

Showing 2 changed files with 23 additions and 10 deletions.
diff --git a/responsibleai_vision/responsibleai_vision/utils/feature_extractors.py b/responsibleai_vision/responsibleai_vision/utils/feature_extractors.py
@@ -80,7 +80,7 @@ def extract_features(image_dataset: pd.DataFrame,
                 exifdata = im.getexif()
                 for tag_id in exifdata:
                     # get the tag name, instead of human unreadable tag id
-                    tag = TAGS.get(tag_id, tag_id)
+                    tag = str(TAGS.get(tag_id, tag_id))
                     data = exifdata.get(tag_id)
                     # decode bytes
                     if isinstance(data, bytes):
@@ -92,8 +92,8 @@ def extract_features(image_dataset: pd.DataFrame,
                             feature_metadata = FeatureMetadata()
                             feature_metadata.categorical_features = []
                         if tag in feature_names:
-                            feature_metadata.categorical_features.append(
-                                str(tag))
+                            if tag not in feature_metadata.categorical_features:
+                                feature_metadata.categorical_features.append(tag)
                             tag_index = feature_names.index(tag)
                             row_feature_values[tag_index] = data
                         else:

diff --git a/responsibleai_vision/tests/test_feature_extractors.py b/responsibleai_vision/tests/test_feature_extractors.py
@@ -8,6 +8,7 @@
 from responsibleai_vision.common.constants import (ExtractedFeatures,
                                                    ImageColumns, ImageModes)
 from responsibleai_vision.utils.feature_extractors import extract_features
+from responsibleai.feature_metadata import FeatureMetadata
 
 MEAN_PIXEL_VALUE = ExtractedFeatures.MEAN_PIXEL_VALUE.value
 FRIDGE_METADATA_FEATURES = [
@@ -18,17 +19,23 @@
 
 
 def validate_extracted_features(extracted_features, feature_names,
-                                expected_feature_names, data):
+                                expected_feature_names, data,
+                                feature_metadata=None):
     assert len(extracted_features) == len(data)
     assert feature_names[0] == expected_feature_names[0]
     for i in range(1, len(feature_names)):
         assert feature_names[i] in expected_feature_names
     assert len(feature_names) == len(expected_feature_names)
     assert len(extracted_features[0]) == len(feature_names)
+    if feature_metadata is not None:
+        assert len(feature_metadata.categorical_features) <= len(feature_names)
+        for categorical_feature in feature_metadata.categorical_features:
+            assert categorical_feature in feature_names
 
 
-def extract_dataset_features(data):
-    return extract_features(data, ImageColumns.LABEL, ImageModes.RGB, None)
+def extract_dataset_features(data, feature_metadata=None):
+    return extract_features(data, ImageColumns.LABEL, ImageModes.RGB,
+                            feature_metadata=feature_metadata)
 
 
 class TestFeatureExtractors(object):
@@ -55,15 +62,21 @@ def test_extract_features_imagenet_metadata(self):
 
     def test_extract_features_flowers_metadata(self):
         data = load_flowers_dataset(upscale=False)
-        extracted_features, feature_names = extract_dataset_features(data)
+        feature_metadata = FeatureMetadata()
+        extracted_features, feature_names = extract_dataset_features(
+            data, feature_metadata=feature_metadata)
         expected_feature_names = [MEAN_PIXEL_VALUE]
         validate_extracted_features(extracted_features, feature_names,
-                                    expected_feature_names, data)
+                                    expected_feature_names, data,
+                                    feature_metadata)
 
     def test_extract_features_mixed_exif_XPComment_metadata(self):
         data = load_fridge_dataset(add_extra_mixed_metadata=True)
-        extracted_features, feature_names = extract_dataset_features(data)
+        feature_metadata = FeatureMetadata()
+        extracted_features, feature_names = extract_dataset_features(
+            data, feature_metadata=feature_metadata)
         expected_feature_names = [MEAN_PIXEL_VALUE, 'XPComment']
         expected_feature_names += FRIDGE_METADATA_FEATURES
         validate_extracted_features(extracted_features, feature_names,
-                                    expected_feature_names, data)
+                                    expected_feature_names, data,
+                                    feature_metadata)