julien.blanchon
add app
c8c12e9
"""Custom Folder Dataset.
This script creates a custom dataset from a folder.
"""
# Copyright (C) 2020 Intel Corporation
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing,
# software distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions
# and limitations under the License.
import logging
import warnings
from pathlib import Path
from typing import Dict, Optional, Tuple, Union
import albumentations as A
import cv2
import numpy as np
from pandas.core.frame import DataFrame
from pytorch_lightning.core.datamodule import LightningDataModule
from pytorch_lightning.utilities.types import EVAL_DATALOADERS, TRAIN_DATALOADERS
from torch import Tensor
from torch.utils.data import DataLoader, Dataset
from torchvision.datasets.folder import IMG_EXTENSIONS
from anomalib.data.inference import InferenceDataset
from anomalib.data.utils import read_image
from anomalib.data.utils.split import (
create_validation_set_from_test_set,
split_normal_images_in_train_set,
)
from anomalib.pre_processing import PreProcessor
logger = logging.getLogger(__name__)
def _check_and_convert_path(path: Union[str, Path]) -> Path:
"""Check an input path, and convert to Pathlib object.
Args:
path (Union[str, Path]): Input path.
Returns:
Path: Output path converted to pathlib object.
"""
if not isinstance(path, Path):
path = Path(path)
return path
def _prepare_files_labels(
path: Union[str, Path], path_type: str, extensions: Optional[Tuple[str, ...]] = None
) -> Tuple[list, list]:
"""Return a list of filenames and list corresponding labels.
Args:
path (Union[str, Path]): Path to the directory containing images.
path_type (str): Type of images in the provided path ("normal", "abnormal", "normal_test")
extensions (Optional[Tuple[str, ...]], optional): Type of the image extensions to read from the
directory.
Returns:
List, List: Filenames of the images provided in the paths, labels of the images provided in the paths
"""
path = _check_and_convert_path(path)
if extensions is None:
extensions = IMG_EXTENSIONS
if isinstance(extensions, str):
extensions = (extensions,)
filenames = [f for f in path.glob(r"**/*") if f.suffix in extensions and not f.is_dir()]
if len(filenames) == 0:
raise RuntimeError(f"Found 0 {path_type} images in {path}")
labels = [path_type] * len(filenames)
return filenames, labels
def make_dataset(
normal_dir: Union[str, Path],
abnormal_dir: Union[str, Path],
normal_test_dir: Optional[Union[str, Path]] = None,
mask_dir: Optional[Union[str, Path]] = None,
split: Optional[str] = None,
split_ratio: float = 0.2,
seed: int = 0,
create_validation_set: bool = True,
extensions: Optional[Tuple[str, ...]] = None,
):
"""Make Folder Dataset.
Args:
normal_dir (Union[str, Path]): Path to the directory containing normal images.
abnormal_dir (Union[str, Path]): Path to the directory containing abnormal images.
normal_test_dir (Optional[Union[str, Path]], optional): Path to the directory containing
normal images for the test dataset. Normal test images will be a split of `normal_dir`
if `None`. Defaults to None.
mask_dir (Optional[Union[str, Path]], optional): Path to the directory containing
the mask annotations. Defaults to None.
split (Optional[str], optional): Dataset split (ie., either train or test). Defaults to None.
split_ratio (float, optional): Ratio to split normal training images and add to the
test set in case test set doesn't contain any normal images.
Defaults to 0.2.
seed (int, optional): Random seed to ensure reproducibility when splitting. Defaults to 0.
create_validation_set (bool, optional):Boolean to create a validation set from the test set.
Those wanting to create a validation set could set this flag to ``True``.
extensions (Optional[Tuple[str, ...]], optional): Type of the image extensions to read from the
directory.
Returns:
DataFrame: an output dataframe containing samples for the requested split (ie., train or test)
"""
filenames = []
labels = []
dirs = {"normal": normal_dir, "abnormal": abnormal_dir}
if normal_test_dir:
dirs = {**dirs, **{"normal_test": normal_test_dir}}
for dir_type, path in dirs.items():
filename, label = _prepare_files_labels(path, dir_type, extensions)
filenames += filename
labels += label
samples = DataFrame({"image_path": filenames, "label": labels})
# Create label index for normal (0) and abnormal (1) images.
samples.loc[(samples.label == "normal") | (samples.label == "normal_test"), "label_index"] = 0
samples.loc[(samples.label == "abnormal"), "label_index"] = 1
samples.label_index = samples.label_index.astype(int)
# If a path to mask is provided, add it to the sample dataframe.
if mask_dir is not None:
mask_dir = _check_and_convert_path(mask_dir)
samples["mask_path"] = ""
for index, row in samples.iterrows():
if row.label_index == 1:
samples.loc[index, "mask_path"] = str(mask_dir / row.image_path.name)
# Ensure the pathlib objects are converted to str.
# This is because torch dataloader doesn't like pathlib.
samples = samples.astype({"image_path": "str"})
# Create train/test split.
# By default, all the normal samples are assigned as train.
# and all the abnormal samples are test.
samples.loc[(samples.label == "normal"), "split"] = "train"
samples.loc[(samples.label == "abnormal") | (samples.label == "normal_test"), "split"] = "test"
if not normal_test_dir:
samples = split_normal_images_in_train_set(
samples=samples, split_ratio=split_ratio, seed=seed, normal_label="normal"
)
# If `create_validation_set` is set to True, the test set is split into half.
if create_validation_set:
samples = create_validation_set_from_test_set(samples, seed=seed, normal_label="normal")
# Get the data frame for the split.
if split is not None and split in ["train", "val", "test"]:
samples = samples[samples.split == split]
samples = samples.reset_index(drop=True)
return samples
class FolderDataset(Dataset):
"""Folder Dataset."""
def __init__(
self,
normal_dir: Union[Path, str],
abnormal_dir: Union[Path, str],
split: str,
pre_process: PreProcessor,
normal_test_dir: Optional[Union[Path, str]] = None,
split_ratio: float = 0.2,
mask_dir: Optional[Union[Path, str]] = None,
extensions: Optional[Tuple[str, ...]] = None,
task: Optional[str] = None,
seed: int = 0,
create_validation_set: bool = False,
) -> None:
"""Create Folder Folder Dataset.
Args:
normal_dir (Union[str, Path]): Path to the directory containing normal images.
abnormal_dir (Union[str, Path]): Path to the directory containing abnormal images.
split (Optional[str], optional): Dataset split (ie., either train or test). Defaults to None.
pre_process (Optional[PreProcessor], optional): Image Pro-processor to apply transform.
Defaults to None.
normal_test_dir (Optional[Union[str, Path]], optional): Path to the directory containing
normal images for the test dataset. Defaults to None.
split_ratio (float, optional): Ratio to split normal training images and add to the
test set in case test set doesn't contain any normal images.
Defaults to 0.2.
mask_dir (Optional[Union[str, Path]], optional): Path to the directory containing
the mask annotations. Defaults to None.
extensions (Optional[Tuple[str, ...]], optional): Type of the image extensions to read from the
directory.
task (Optional[str], optional): Task type. (classification or segmentation) Defaults to None.
seed (int, optional): Random seed to ensure reproducibility when splitting. Defaults to 0.
create_validation_set (bool, optional):Boolean to create a validation set from the test set.
Those wanting to create a validation set could set this flag to ``True``.
Raises:
ValueError: When task is set to classification and `mask_dir` is provided. When `mask_dir` is
provided, `task` should be set to `segmentation`.
"""
self.split = split
if task == "segmentation" and mask_dir is None:
warnings.warn(
"Segmentation task is requested, but mask directory is not provided. "
"Classification is to be chosen if mask directory is not provided."
)
self.task = "classification"
if task == "classification" and mask_dir:
warnings.warn(
"Classification task is requested, but mask directory is provided. "
"Segmentation task is to be chosen if mask directory is provided."
)
self.task = "segmentation"
if task is None or mask_dir is None:
self.task = "classification"
else:
self.task = task
self.pre_process = pre_process
self.samples = make_dataset(
normal_dir=normal_dir,
abnormal_dir=abnormal_dir,
normal_test_dir=normal_test_dir,
mask_dir=mask_dir,
split=split,
split_ratio=split_ratio,
seed=seed,
create_validation_set=create_validation_set,
extensions=extensions,
)
def __len__(self) -> int:
"""Get length of the dataset."""
return len(self.samples)
def __getitem__(self, index: int) -> Dict[str, Union[str, Tensor]]:
"""Get dataset item for the index ``index``.
Args:
index (int): Index to get the item.
Returns:
Union[Dict[str, Tensor], Dict[str, Union[str, Tensor]]]: Dict of image tensor during training.
Otherwise, Dict containing image path, target path, image tensor, label and transformed bounding box.
"""
item: Dict[str, Union[str, Tensor]] = {}
image_path = self.samples.image_path[index]
image = read_image(image_path)
pre_processed = self.pre_process(image=image)
item = {"image": pre_processed["image"]}
if self.split in ["val", "test"]:
label_index = self.samples.label_index[index]
item["image_path"] = image_path
item["label"] = label_index
if self.task == "segmentation":
mask_path = self.samples.mask_path[index]
# Only Anomalous (1) images has masks in MVTec AD dataset.
# Therefore, create empty mask for Normal (0) images.
if label_index == 0:
mask = np.zeros(shape=image.shape[:2])
else:
mask = cv2.imread(mask_path, flags=0) / 255.0
pre_processed = self.pre_process(image=image, mask=mask)
item["mask_path"] = mask_path
item["image"] = pre_processed["image"]
item["mask"] = pre_processed["mask"]
return item
class FolderDataModule(LightningDataModule):
"""Folder Lightning Data Module."""
def __init__(
self,
root: Union[str, Path],
normal_dir: str = "normal",
abnormal_dir: str = "abnormal",
task: str = "classification",
normal_test_dir: Optional[Union[Path, str]] = None,
mask_dir: Optional[Union[Path, str]] = None,
extensions: Optional[Tuple[str, ...]] = None,
split_ratio: float = 0.2,
seed: int = 0,
image_size: Optional[Union[int, Tuple[int, int]]] = None,
train_batch_size: int = 32,
test_batch_size: int = 32,
num_workers: int = 8,
transform_config_train: Optional[Union[str, A.Compose]] = None,
transform_config_val: Optional[Union[str, A.Compose]] = None,
create_validation_set: bool = False,
) -> None:
"""Folder Dataset PL Datamodule.
Args:
root (Union[str, Path]): Path to the root folder containing normal and abnormal dirs.
normal_dir (str, optional): Name of the directory containing normal images.
Defaults to "normal".
abnormal_dir (str, optional): Name of the directory containing abnormal images.
Defaults to "abnormal".
task (str, optional): Task type. Could be either classification or segmentation.
Defaults to "classification".
normal_test_dir (Optional[Union[str, Path]], optional): Path to the directory containing
normal images for the test dataset. Defaults to None.
mask_dir (Optional[Union[str, Path]], optional): Path to the directory containing
the mask annotations. Defaults to None.
extensions (Optional[Tuple[str, ...]], optional): Type of the image extensions to read from the
directory. Defaults to None.
split_ratio (float, optional): Ratio to split normal training images and add to the
test set in case test set doesn't contain any normal images.
Defaults to 0.2.
seed (int, optional): Random seed to ensure reproducibility when splitting. Defaults to 0.
image_size (Optional[Union[int, Tuple[int, int]]], optional): Size of the input image.
Defaults to None.
train_batch_size (int, optional): Training batch size. Defaults to 32.
test_batch_size (int, optional): Test batch size. Defaults to 32.
num_workers (int, optional): Number of workers. Defaults to 8.
transform_config_train (Optional[Union[str, A.Compose]], optional): Config for pre-processing
during training.
Defaults to None.
transform_config_val (Optional[Union[str, A.Compose]], optional): Config for pre-processing
during validation.
Defaults to None.
create_validation_set (bool, optional):Boolean to create a validation set from the test set.
Those wanting to create a validation set could set this flag to ``True``.
Examples:
Assume that we use Folder Dataset for the MVTec/bottle/broken_large category. We would do:
>>> from anomalib.data import FolderDataModule
>>> datamodule = FolderDataModule(
... root="./datasets/MVTec/bottle/test",
... normal="good",
... abnormal="broken_large",
... image_size=256
... )
>>> datamodule.setup()
>>> i, data = next(enumerate(datamodule.train_dataloader()))
>>> data["image"].shape
torch.Size([16, 3, 256, 256])
>>> i, test_data = next(enumerate(datamodule.test_dataloader()))
>>> test_data.keys()
dict_keys(['image'])
We could also create a Folder DataModule for datasets containing mask annotations.
The dataset expects that mask annotation filenames must be same as the original filename.
To this end, we modified mask filenames in MVTec AD bottle category.
Now we could try folder data module using the mvtec bottle broken large category
>>> datamodule = FolderDataModule(
... root="./datasets/bottle/test",
... normal="good",
... abnormal="broken_large",
... mask_dir="./datasets/bottle/ground_truth/broken_large",
... image_size=256
... )
>>> i , train_data = next(enumerate(datamodule.train_dataloader()))
>>> train_data.keys()
dict_keys(['image'])
>>> train_data["image"].shape
torch.Size([16, 3, 256, 256])
>>> i, test_data = next(enumerate(datamodule.test_dataloader()))
dict_keys(['image_path', 'label', 'mask_path', 'image', 'mask'])
>>> print(test_data["image"].shape, test_data["mask"].shape)
torch.Size([24, 3, 256, 256]) torch.Size([24, 256, 256])
By default, Folder Data Module does not create a validation set. If a validation set
is needed it could be set as follows:
>>> datamodule = FolderDataModule(
... root="./datasets/bottle/test",
... normal="good",
... abnormal="broken_large",
... mask_dir="./datasets/bottle/ground_truth/broken_large",
... image_size=256,
... create_validation_set=True,
... )
>>> i, val_data = next(enumerate(datamodule.val_dataloader()))
>>> val_data.keys()
dict_keys(['image_path', 'label', 'mask_path', 'image', 'mask'])
>>> print(val_data["image"].shape, val_data["mask"].shape)
torch.Size([12, 3, 256, 256]) torch.Size([12, 256, 256])
>>> i, test_data = next(enumerate(datamodule.test_dataloader()))
>>> print(test_data["image"].shape, test_data["mask"].shape)
torch.Size([12, 3, 256, 256]) torch.Size([12, 256, 256])
"""
super().__init__()
self.root = _check_and_convert_path(root)
self.normal_dir = self.root / normal_dir
self.abnormal_dir = self.root / abnormal_dir
self.normal_test = normal_test_dir
if normal_test_dir:
self.normal_test = self.root / normal_test_dir
self.mask_dir = mask_dir
self.extensions = extensions
self.split_ratio = split_ratio
if task == "classification" and mask_dir is not None:
raise ValueError(
"Classification type is set but mask_dir provided. "
"If mask_dir is provided task type must be segmentation. "
"Check your configuration."
)
self.task = task
self.transform_config_train = transform_config_train
self.transform_config_val = transform_config_val
self.image_size = image_size
if self.transform_config_train is not None and self.transform_config_val is None:
self.transform_config_val = self.transform_config_train
self.pre_process_train = PreProcessor(config=self.transform_config_train, image_size=self.image_size)
self.pre_process_val = PreProcessor(config=self.transform_config_val, image_size=self.image_size)
self.train_batch_size = train_batch_size
self.test_batch_size = test_batch_size
self.num_workers = num_workers
self.create_validation_set = create_validation_set
self.seed = seed
self.train_data: Dataset
self.test_data: Dataset
if create_validation_set:
self.val_data: Dataset
self.inference_data: Dataset
def setup(self, stage: Optional[str] = None) -> None:
"""Setup train, validation and test data.
Args:
stage: Optional[str]: Train/Val/Test stages. (Default value = None)
"""
logger.info("Setting up train, validation, test and prediction datasets.")
if stage in (None, "fit"):
self.train_data = FolderDataset(
normal_dir=self.normal_dir,
abnormal_dir=self.abnormal_dir,
normal_test_dir=self.normal_test,
split="train",
split_ratio=self.split_ratio,
mask_dir=self.mask_dir,
pre_process=self.pre_process_train,
extensions=self.extensions,
task=self.task,
seed=self.seed,
create_validation_set=self.create_validation_set,
)
if self.create_validation_set:
self.val_data = FolderDataset(
normal_dir=self.normal_dir,
abnormal_dir=self.abnormal_dir,
normal_test_dir=self.normal_test,
split="val",
split_ratio=self.split_ratio,
mask_dir=self.mask_dir,
pre_process=self.pre_process_val,
extensions=self.extensions,
task=self.task,
seed=self.seed,
create_validation_set=self.create_validation_set,
)
self.test_data = FolderDataset(
normal_dir=self.normal_dir,
abnormal_dir=self.abnormal_dir,
split="test",
normal_test_dir=self.normal_test,
split_ratio=self.split_ratio,
mask_dir=self.mask_dir,
pre_process=self.pre_process_val,
extensions=self.extensions,
task=self.task,
seed=self.seed,
create_validation_set=self.create_validation_set,
)
if stage == "predict":
self.inference_data = InferenceDataset(
path=self.root, image_size=self.image_size, transform_config=self.transform_config_val
)
def train_dataloader(self) -> TRAIN_DATALOADERS:
"""Get train dataloader."""
return DataLoader(self.train_data, shuffle=True, batch_size=self.train_batch_size, num_workers=self.num_workers)
def val_dataloader(self) -> EVAL_DATALOADERS:
"""Get validation dataloader."""
dataset = self.val_data if self.create_validation_set else self.test_data
return DataLoader(dataset=dataset, shuffle=False, batch_size=self.test_batch_size, num_workers=self.num_workers)
def test_dataloader(self) -> EVAL_DATALOADERS:
"""Get test dataloader."""
return DataLoader(self.test_data, shuffle=False, batch_size=self.test_batch_size, num_workers=self.num_workers)
def predict_dataloader(self) -> EVAL_DATALOADERS:
"""Get predict dataloader."""
return DataLoader(
self.inference_data, shuffle=False, batch_size=self.test_batch_size, num_workers=self.num_workers
)