Support multiple labels for kaggle format

CHANGELOG.md

@@ -5,7 +5,16 @@ All notable changes to this project will be documented in this file.
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
-## \[Q3 2024 Release 1.9.0\]
+## \[Q4 2024 Release 1.9.1\]
+### New features
+
+### Enhancements
+- Support multiple labels for kaggle format
+  (<https://github.com/openvinotoolkit/datumaro/pull/1607>)
+
+### Bug fixes
+
+## Q3 2024 Release 1.9.0
 ### New features
 - Add a new CLI command: datum format
   (<https://github.com/openvinotoolkit/datumaro/pull/1570>)

src/datumaro/plugins/data_formats/kaggle/base.py

@@ -77,21 +77,43 @@ def _parse_bbox_coords(self, bbox_str):
         # expected to output [x1, y1, x2, y2]
         return [float(coord.strip()) for coord in coords]
 
-    def _load_annotations(self, datas: list, indices: Dict[str, int], bbox_flag: bool):
+    def _load_annotations(
+        self, datas: list, indices: Dict[str, Union[int, Dict[str, int]]], bbox_flag: bool
+    ):
         if "label" in indices:
-            label_name = str(datas[indices["label"]])
-            label, cat = self._label_cat.find(label_name)
-            if not cat:
-                self._label_cat.add(label_name)
-                label, _ = self._label_cat.find(label_name)
+            label_indices = indices["label"]
+            if isinstance(label_indices, dict):
+                labels = []
+                list_values = datas[1:]
+                index_to_label = {v: k for k, v in label_indices.items()}
+                present_labels = [
+                    index_to_label[i + 1] for i, value in enumerate(list_values) if value == "1"
+                ]
+
+                for label_name in present_labels:
+                    label, cat = self._label_cat.find(label_name)
+                    if not cat:
+                        self._label_cat.add(label_name)
+                        label, _ = self._label_cat.find(label_name)
+                    labels.append(Label(label=label))
+            else:
+                label_name = str(datas[indices["label"]])
+                label, cat = self._label_cat.find(label_name)
+                if not cat:
+                    self._label_cat.add(label_name)
+                    label, _ = self._label_cat.find(label_name)
         else:
             _, cat = self._label_cat.find("object")
             if not cat:
                 self._label_cat.add("object")
             label = 0
 
         if "label" in indices and not bbox_flag:
+            label_indices = indices["label"]
+            if isinstance(label_indices, dict):
+                return labels
             return Label(label=label)
+
         if bbox_flag:
             if "bbox" in indices:
                 coords = self._parse_bbox_coords(datas[indices["bbox"]])
@@ -125,7 +147,14 @@ def _load_items(self, ann_file: str, columns: Dict[str, Union[str, list]]):
 
         indices = {"media": df_fields.index(columns["media"])}
         if "label" in columns:
-            indices.update({"label": df_fields.index(columns["label"])})
+            label_columns = columns["label"]
+            if isinstance(label_columns, list):
+                indices_label = {}
+                for label in label_columns:
+                    indices_label[label] = df_fields.index(label)
+                indices.update({"label": indices_label})
+            else:
+                indices.update({"label": df_fields.index(label_columns)})
 
         bbox_flag = False
         bbox_index = columns.get("bbox")
@@ -165,16 +194,30 @@ def _load_items(self, ann_file: str, columns: Dict[str, Union[str, list]]):
                 continue
 
             ann = self._load_annotations(data_info, indices, bbox_flag)
-            self._ann_types.add(ann.type)
-            if item_id in items:
-                items[item_id].annotations.append(ann)
+            if isinstance(ann, list):
+                for label in ann:
+                    self._ann_types.add(label.type)
+                if item_id in items:
+                    for label in ann:
+                        items[item_id].annotations.append(label)
+                else:
+                    items[item_id] = DatasetItem(
+                        id=item_id,
+                        subset=self._subset,
+                        media=Image.from_file(path=media_path),
+                        annotations=ann,
+                    )
             else:
-                items[item_id] = DatasetItem(
-                    id=item_id,
-                    subset=self._subset,
-                    media=Image.from_file(path=media_path),
-                    annotations=[ann],
-                )
+                self._ann_types.add(ann.type)
+                if item_id in items:
+                    items[item_id].annotations.append(ann)
+                else:
+                    items[item_id] = DatasetItem(
+                        id=item_id,
+                        subset=self._subset,
+                        media=Image.from_file(path=media_path),
+                        annotations=[ann],
+                    )
         return items.values()
 
     def categories(self):

tests/assets/kaggle_dataset/image_csv_multi_label/ann.csv

@@ -0,0 +1,7 @@
+image_name,dog,cat,person
+1.jpg,1,0,0
+2.jpg,0,1,0
+3.jpg,0,0,1
+4.jpg,1,1,0
+5.jpg,1,0,1
+6.jpg,0,1,1

tests/assets/kaggle_dataset/image_csv_multi_label/ann_wo_ext.csv

@@ -0,0 +1,7 @@
+image_name,dog,cat,person
+1,1,0,0
+2,0,1,0
+3,0,0,1
+4,1,1,0
+5,1,0,1
+6,0,1,1

tests/unit/data_formats/test_kaggle.py

@@ -20,6 +20,9 @@
 from tests.utils.test_utils import compare_datasets
 
 DUMMY_DATASET_IMAGE_CSV_DIR = get_test_asset_path("kaggle_dataset", "image_csv")
+DUMMY_DATASET_IMAGE_CSV_MULTI_LB_DIR = get_test_asset_path(
+    "kaggle_dataset", "image_csv_multi_label"
+)
 DUMMY_DATASET_IMAGE_CSV_DET_DIR = get_test_asset_path("kaggle_dataset", "image_csv_det")
 DUMMY_DATASET_IMAGE_TXT_DIR = get_test_asset_path("kaggle_dataset", "image_txt")
 DUMMY_DATASET_IMAGE_TXT_DET_DIR = get_test_asset_path("kaggle_dataset", "image_txt_det")
@@ -72,6 +75,51 @@ def fxt_img_dataset() -> Dataset:
     )
 
 
+@pytest.fixture
+def fxt_img_multi_label_dataset() -> Dataset:
+    return Dataset.from_iterable(
+        [
+            DatasetItem(
+                id="1",
+                subset="default",
+                media=Image.from_numpy(data=np.ones((5, 10, 3))),
+                annotations=[Label(label=0)],
+            ),
+            DatasetItem(
+                id="2",
+                subset="default",
+                media=Image.from_numpy(data=np.ones((5, 10, 3))),
+                annotations=[Label(label=1)],
+            ),
+            DatasetItem(
+                id="3",
+                subset="default",
+                media=Image.from_numpy(data=np.ones((5, 10, 3))),
+                annotations=[Label(label=2)],
+            ),
+            DatasetItem(
+                id="4",
+                subset="default",
+                media=Image.from_numpy(data=np.ones((5, 10, 3))),
+                annotations=[Label(label=0), Label(label=1)],
+            ),
+            DatasetItem(
+                id="5",
+                subset="default",
+                media=Image.from_numpy(data=np.ones((5, 10, 3))),
+                annotations=[Label(label=0), Label(label=2)],
+            ),
+            DatasetItem(
+                id="6",
+                subset="default",
+                media=Image.from_numpy(data=np.ones((5, 10, 3))),
+                annotations=[Label(label=1), Label(label=2)],
+            ),
+        ],
+        categories=["dog", "cat", "person"],
+    )
+
+
 @pytest.fixture
 def fxt_img_det_dataset() -> Dataset:
     return Dataset.from_iterable(
@@ -321,6 +369,8 @@ def fxt_coco_dataset() -> Dataset:
 IDS = [
     "IMAGE_CSV",
     "IMAGE_CSV_WO_EXT",
+    "IMAGE_CSV_MULTI_LB",
+    "IMAGE_CSV_MULTI_LB_WO_EXT",
     "IMAGE_CSV_DET",
     "IMAGE_CSV_DET2",
     "IMAGE_CSV_DET3",
@@ -372,6 +422,26 @@ def test_can_detect(self, fxt_dataset_dir: str):
                     "columns": {"media": "image_name", "label": "label_name"},
                 },
             ),
+            (
+                DUMMY_DATASET_IMAGE_CSV_MULTI_LB_DIR,
+                "images",
+                "fxt_img_multi_label_dataset",
+                KaggleImageCsvBase,
+                {
+                    "ann_file": osp.join(DUMMY_DATASET_IMAGE_CSV_MULTI_LB_DIR, "ann.csv"),
+                    "columns": {"media": "image_name", "label": ["dog", "cat", "person"]},
+                },
+            ),
+            (
+                DUMMY_DATASET_IMAGE_CSV_MULTI_LB_DIR,
+                "images",
+                "fxt_img_multi_label_dataset",
+                KaggleImageCsvBase,
+                {
+                    "ann_file": osp.join(DUMMY_DATASET_IMAGE_CSV_MULTI_LB_DIR, "ann_wo_ext.csv"),
+                    "columns": {"media": "image_name", "label": ["dog", "cat", "person"]},
+                },
+            ),
             (
                 DUMMY_DATASET_IMAGE_CSV_DET_DIR,
                 "images",