Update the tests

Signed-off-by: Samet Akcay <[email protected]>

Author: samet-akcay

src/anomalib/data/datamodules/image/datumaro.py

@@ -11,7 +11,7 @@
 from anomalib import TaskType
 from anomalib.data.datamodules.base import AnomalibDataModule
 from anomalib.data.datasets.image.datumaro import DatumaroDataset
-from anomalib.data.utils import Split, TestSplitMode, ValSplitMode
+from anomalib.data.utils import Split, SplitMode, TestSplitMode, ValSplitMode
 
 
 class Datumaro(AnomalibDataModule):
@@ -69,9 +69,9 @@ def __init__(
         eval_batch_size: int = 32,
         num_workers: int = 8,
         task: TaskType = TaskType.CLASSIFICATION,
-        test_split_mode: TestSplitMode | str = TestSplitMode.FROM_DIR,
-        test_split_ratio: float = 0.5,
-        val_split_mode: ValSplitMode | str = ValSplitMode.FROM_TEST,
+        test_split_mode: SplitMode | TestSplitMode | str = SplitMode.AUTO,
+        test_split_ratio: float | None = None,
+        val_split_mode: SplitMode | ValSplitMode | str = SplitMode.AUTO,
         val_split_ratio: float = 0.5,
         seed: int | None = None,
     ) -> None:

src/anomalib/data/datasets/base/image.py

@@ -288,6 +288,7 @@ def __getitem__(self, index: int) -> DatasetItem:
             DatasetItem: DatasetItem instance containing image and ground truth (if available).
         """
         image_path = self.samples.iloc[index].image_path
+        mask_path = self.samples.iloc[index].get("mask_path") if "mask_path" in self.samples.columns else None
         label_index = self.samples.iloc[index].label_index
 
         image = read_image(image_path, as_tensor=True)
@@ -296,16 +297,17 @@ def __getitem__(self, index: int) -> DatasetItem:
         if self.task == TaskType.CLASSIFICATION:
             item["image"] = self.transform(image) if self.transform else image
         elif self.task == TaskType.SEGMENTATION:
+            if mask_path is None:
+                msg = "mask_path is required for segmentation tasks but was not found in samples DataFrame"
+                raise ValueError(msg)
             # Only Anomalous (1) images have masks in anomaly datasets
             # Therefore, create empty mask for Normal (0) images.
-            mask_path = self.samples.iloc[index].mask_path
             mask = (
                 Mask(torch.zeros(image.shape[-2:])).to(torch.uint8)
                 if label_index == LabelName.NORMAL
                 else read_mask(mask_path, as_tensor=True)
             )
             item["image"], item["gt_mask"] = self.transform(image, mask) if self.transform else (image, mask)
-
         else:
             msg = f"Unknown task type: {self.task}"
             raise ValueError(msg)

src/anomalib/data/datasets/base/video.py

@@ -4,8 +4,8 @@
 # SPDX-License-Identifier: Apache-2.0
 
 from abc import ABC
+from collections.abc import Callable
 from enum import Enum
-from typing import TYPE_CHECKING
 
 import torch
 from pandas import DataFrame
@@ -14,14 +14,11 @@
 from torchvision.tv_tensors import Mask
 
 from anomalib import TaskType
-from anomalib.data.dataclasses import VideoItem
+from anomalib.data.dataclasses import VideoBatch, VideoItem
 from anomalib.data.utils.video import ClipsIndexer
 
 from .image import AnomalibDataset
 
-if TYPE_CHECKING:
-    from collections.abc import Callable
-
 
 class VideoTargetFrame(str, Enum):
     """Target frame for a video-clip.
@@ -172,3 +169,8 @@ def __getitem__(self, index: int) -> VideoItem:
             item = self._select_targets(item)
 
         return item
+
+    @property
+    def collate_fn(self) -> Callable:
+        """Return the collate function for video batches."""
+        return VideoBatch.collate

src/anomalib/data/datasets/image/kolektor.py

@@ -17,7 +17,6 @@
 # Copyright (C) 2024 Intel Corporation
 # SPDX-License-Identifier: Apache-2.0
 
-import logging
 from pathlib import Path
 
 import numpy as np
@@ -29,19 +28,10 @@
 from anomalib import TaskType
 from anomalib.data.datasets import AnomalibDataset
 from anomalib.data.errors import MisMatchError
-from anomalib.data.utils import DownloadInfo, Split, validate_path
+from anomalib.data.utils import Split, validate_path
 
 __all__ = ["KolektorDataset", "make_kolektor_dataset"]
 
-logger = logging.getLogger(__name__)
-
-DOWNLOAD_INFO = DownloadInfo(
-    name="kolektor",
-    url="https://go.vicos.si/kolektorsdd",
-    hashsum="65dc621693418585de9c4467d1340ea7958a6181816f0dc2883a1e8b61f9d4dc",
-    filename="KolektorSDD.zip",
-)
-
 
 class KolektorDataset(AnomalibDataset):
     """Kolektor dataset class.

tests/helpers/data.py

@@ -265,8 +265,8 @@ class DummyImageDatasetGenerator(DummyDatasetGenerator):
     Args:
         data_format (DataFormat): Data format of the dataset.
         root (Path | str, optional): Path to the root directory. Defaults to None.
-        num_train (int, optional): Number of training images to generate. Defaults to 5.
-        num_test (int, optional): Number of testing images to generate per category. Defaults to 5.
+        num_train (int, optional): Number of training images to generate. Defaults to 8.
+        num_test (int, optional): Number of testing images to generate per category. Defaults to 8.
         img_height (int, optional): Height of the image. Defaults to 128.
         img_width (int, optional): Width of the image. Defaults to 128.
         max_size (Optional[int], optional): Maximum size of the test shapes. Defaults to 10.
@@ -301,8 +301,8 @@ def __init__(
         root: Path | str | None = None,
         normal_category: str = "good",
         abnormal_category: str = "bad",
-        num_train: int = 5,
-        num_test: int = 5,
+        num_train: int = 8,
+        num_test: int = 8,
         image_shape: tuple[int, int] = (256, 256),
         num_channels: int = 3,
         min_size: int = 64,

tests/unit/data/datamodule/image/test_csv.py

@@ -26,8 +26,8 @@ def datamodule(dataset_path: Path, task_type: TaskType) -> CSV:
         _datamodule = CSV(
             name="dummy_csv",
             csv_path=dataset_path / "csv" / "samples.csv",
-            train_batch_size=2,
-            eval_batch_size=2,
+            train_batch_size=4,
+            eval_batch_size=4,
             num_workers=0,
             task=task_type,
             test_split_mode="predefined",

tests/unit/data/datamodule/image/test_folder.py

@@ -33,8 +33,8 @@ def datamodule(dataset_path: Path, task_type: TaskType) -> Folder:
             normal_dir="train/good",
             abnormal_dir="test/bad",
             mask_dir=mask_dir,
-            train_batch_size=2,
-            eval_batch_size=2,
+            train_batch_size=4,
+            eval_batch_size=4,
             num_workers=0,
             task=task_type,
         )

tests/unit/data/datamodule/image/test_mvtec.py

@@ -23,8 +23,8 @@ def datamodule(dataset_path: Path, task_type: TaskType) -> MVTec:
             root=dataset_path / "mvtec",
             category="dummy",
             task=task_type,
-            train_batch_size=2,
-            eval_batch_size=2,
+            train_batch_size=4,
+            eval_batch_size=4,
         )
         _datamodule.prepare_data()
         _datamodule.setup()

tests/unit/data/utils/test_image.py

@@ -25,7 +25,7 @@ def test_existing_image_directory(dataset_path: Path) -> None:
         """Test ``get_image_filenames`` returns the correct image filenames from an existing directory."""
         directory_path = dataset_path / "mvtec/dummy/train/good"
         image_filenames = get_image_filenames(directory_path)
-        expected_filenames = [(directory_path / f"{i:03d}.png").resolve() for i in range(5)]
+        expected_filenames = [(directory_path / f"{i:03d}.png").resolve() for i in range(8)]
         assert set(image_filenames) == set(expected_filenames)
 
     @staticmethod