oodeel 0.2.0__tar.gz → 0.3.0__tar.gz

{oodeel-0.2.0 → oodeel-0.3.0}/PKG-INFO

@@ -1,6 +1,6 @@
-Metadata-Version: 2.1
+Metadata-Version: 2.4
 Name: oodeel
-Version: 0.2.0
+Version: 0.3.0
 Summary: Simple, compact, and hackable post-hoc deep OOD detection for alreadytrained tensorflow or pytorch image classifiers.
 Author: DEEL Core Team
 Author-email: paul.novello@irt-saintexupery.com
@@ -11,13 +11,127 @@ Classifier: Programming Language :: Python :: 3.8
 Classifier: Programming Language :: Python :: 3.9
 Classifier: Programming Language :: Python :: 3.10
 Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: faiss_cpu
+Requires-Dist: numpy
+Requires-Dist: scikit_learn
+Requires-Dist: scipy
+Requires-Dist: setuptools
+Requires-Dist: matplotlib
+Requires-Dist: pandas
+Requires-Dist: seaborn
+Requires-Dist: plotly
+Requires-Dist: tqdm
 Provides-Extra: dev
+Requires-Dist: mypy; extra == "dev"
+Requires-Dist: ipywidgets; extra == "dev"
+Requires-Dist: mkdocs-jupyter; extra == "dev"
+Requires-Dist: mkdocstrings-python; extra == "dev"
+Requires-Dist: flake8; extra == "dev"
+Requires-Dist: setuptools; extra == "dev"
+Requires-Dist: pre-commit; extra == "dev"
+Requires-Dist: tox; extra == "dev"
+Requires-Dist: black; extra == "dev"
+Requires-Dist: ruff; extra == "dev"
+Requires-Dist: ipython; extra == "dev"
+Requires-Dist: ipykernel; extra == "dev"
+Requires-Dist: pytest; extra == "dev"
+Requires-Dist: pylint; extra == "dev"
+Requires-Dist: mypy; extra == "dev"
+Requires-Dist: mkdocs; extra == "dev"
+Requires-Dist: mkdocs-material; extra == "dev"
+Requires-Dist: mkdocstrings; extra == "dev"
+Requires-Dist: mknotebooks; extra == "dev"
+Requires-Dist: mike; extra == "dev"
+Requires-Dist: bump2version; extra == "dev"
+Requires-Dist: docsig; extra == "dev"
+Requires-Dist: no_implicit_optional; extra == "dev"
+Requires-Dist: numpy==1.26.4; extra == "dev"
+Requires-Dist: tensorflow==2.11.0; extra == "dev"
+Requires-Dist: tensorflow_datasets; extra == "dev"
+Requires-Dist: tensorflow_probability==0.19.0; extra == "dev"
+Requires-Dist: timm; extra == "dev"
+Requires-Dist: torch==1.13.1; extra == "dev"
+Requires-Dist: torchvision==0.14.1; extra == "dev"
 Provides-Extra: tensorflow-dev
+Requires-Dist: mypy; extra == "tensorflow-dev"
+Requires-Dist: ipywidgets; extra == "tensorflow-dev"
+Requires-Dist: mkdocs-jupyter; extra == "tensorflow-dev"
+Requires-Dist: mkdocstrings-python; extra == "tensorflow-dev"
+Requires-Dist: flake8; extra == "tensorflow-dev"
+Requires-Dist: setuptools; extra == "tensorflow-dev"
+Requires-Dist: pre-commit; extra == "tensorflow-dev"
+Requires-Dist: tox; extra == "tensorflow-dev"
+Requires-Dist: black; extra == "tensorflow-dev"
+Requires-Dist: ruff; extra == "tensorflow-dev"
+Requires-Dist: ipython; extra == "tensorflow-dev"
+Requires-Dist: ipykernel; extra == "tensorflow-dev"
+Requires-Dist: pytest; extra == "tensorflow-dev"
+Requires-Dist: pylint; extra == "tensorflow-dev"
+Requires-Dist: mypy; extra == "tensorflow-dev"
+Requires-Dist: mkdocs; extra == "tensorflow-dev"
+Requires-Dist: mkdocs-material; extra == "tensorflow-dev"
+Requires-Dist: mkdocstrings; extra == "tensorflow-dev"
+Requires-Dist: mknotebooks; extra == "tensorflow-dev"
+Requires-Dist: mike; extra == "tensorflow-dev"
+Requires-Dist: bump2version; extra == "tensorflow-dev"
+Requires-Dist: docsig; extra == "tensorflow-dev"
+Requires-Dist: no_implicit_optional; extra == "tensorflow-dev"
+Requires-Dist: numpy==1.26.4; extra == "tensorflow-dev"
+Requires-Dist: tensorflow==2.11.0; extra == "tensorflow-dev"
+Requires-Dist: tensorflow_datasets; extra == "tensorflow-dev"
+Requires-Dist: tensorflow_probability==0.19.0; extra == "tensorflow-dev"
 Provides-Extra: torch-dev
+Requires-Dist: mypy; extra == "torch-dev"
+Requires-Dist: ipywidgets; extra == "torch-dev"
+Requires-Dist: mkdocs-jupyter; extra == "torch-dev"
+Requires-Dist: mkdocstrings-python; extra == "torch-dev"
+Requires-Dist: flake8; extra == "torch-dev"
+Requires-Dist: setuptools; extra == "torch-dev"
+Requires-Dist: pre-commit; extra == "torch-dev"
+Requires-Dist: tox; extra == "torch-dev"
+Requires-Dist: black; extra == "torch-dev"
+Requires-Dist: ruff; extra == "torch-dev"
+Requires-Dist: ipython; extra == "torch-dev"
+Requires-Dist: ipykernel; extra == "torch-dev"
+Requires-Dist: pytest; extra == "torch-dev"
+Requires-Dist: pylint; extra == "torch-dev"
+Requires-Dist: mypy; extra == "torch-dev"
+Requires-Dist: mkdocs; extra == "torch-dev"
+Requires-Dist: mkdocs-material; extra == "torch-dev"
+Requires-Dist: mkdocstrings; extra == "torch-dev"
+Requires-Dist: mknotebooks; extra == "torch-dev"
+Requires-Dist: mike; extra == "torch-dev"
+Requires-Dist: bump2version; extra == "torch-dev"
+Requires-Dist: docsig; extra == "torch-dev"
+Requires-Dist: no_implicit_optional; extra == "torch-dev"
+Requires-Dist: numpy==1.26.4; extra == "torch-dev"
+Requires-Dist: timm; extra == "torch-dev"
+Requires-Dist: torch==1.13.1; extra == "torch-dev"
+Requires-Dist: torchvision==0.14.1; extra == "torch-dev"
 Provides-Extra: tensorflow
+Requires-Dist: tensorflow==2.11.0; extra == "tensorflow"
+Requires-Dist: tensorflow_datasets; extra == "tensorflow"
+Requires-Dist: tensorflow_probability==0.19.0; extra == "tensorflow"
 Provides-Extra: torch
+Requires-Dist: timm; extra == "torch"
+Requires-Dist: torch==1.13.1; extra == "torch"
+Requires-Dist: torchvision==0.14.1; extra == "torch"
 Provides-Extra: docs
-License-File: LICENSE
+Requires-Dist: mkdocs; extra == "docs"
+Requires-Dist: mkdocs-material; extra == "docs"
+Requires-Dist: mkdocstrings; extra == "docs"
+Requires-Dist: mknotebooks; extra == "docs"
+Requires-Dist: ipython; extra == "docs"
+Dynamic: author
+Dynamic: author-email
+Dynamic: classifier
+Dynamic: description
+Dynamic: description-content-type
+Dynamic: license-file
+Dynamic: provides-extra
+Dynamic: requires-dist
+Dynamic: summary
 
 
 <!-- Banner section -->
@@ -33,29 +147,23 @@ License-File: LICENSE
 <!-- Badge section -->
 <div align="center">
     <a href="#">
-        <img src="https://img.shields.io/badge/python-3.8%2B-blue">
-    </a>
+        <img src="https://img.shields.io/badge/python-3.8%2B-blue"></a>
     <a href="https://github.com/deel-ai/oodeel/actions/workflows/python-linters.yml">
-        <img alt="Flake8" src="https://github.com/deel-ai/oodeel/actions/workflows/python-linters.yml/badge.svg">
-    </a>
+        <img alt="Flake8" src="https://github.com/deel-ai/oodeel/actions/workflows/python-linters.yml/badge.svg"></a>
     <a href="https://github.com/deel-ai/oodeel/actions/workflows/python-tests-tf.yml">
-        <img alt="Tests tf" src="https://github.com/deel-ai/oodeel/actions/workflows/python-tests-tf.yml/badge.svg">
-    </a>
+        <img alt="Tests tf" src="https://github.com/deel-ai/oodeel/actions/workflows/python-tests-tf.yml/badge.svg"></a>
     <a href="https://github.com/deel-ai/oodeel/actions/workflows/python-tests-torch.yml">
-        <img alt="Tests torch" src="https://github.com/deel-ai/oodeel/actions/workflows/python-tests-torch.yml/badge.svg">
-    </a>
+        <img alt="Tests torch" src="https://github.com/deel-ai/oodeel/actions/workflows/python-tests-torch.yml/badge.svg"></a>
     <a href="https://github.com/deel-ai/oodeel/actions/workflows/python-coverage-shield.yml">
-        <img alt="Coverage" src="https://github.com/deel-ai/oodeel/raw/gh-shields/coverage.svg">
-    </a>
+        <img alt="Coverage" src="https://github.com/deel-ai/oodeel/raw/gh-shields/coverage.svg"></a>
     <a href="https://github.com/deel-ai/oodeel/blob/master/LICENSE">
-        <img alt="License MIT" src="https://img.shields.io/badge/License-MIT-efefef">
-    </a>
+        <img alt="License MIT" src="https://img.shields.io/badge/License-MIT-efefef"></a>
 </div>
 <br>
 
 <!-- Short description of your library -->
 
-<b>Oodeel</b> is a library that performs post-hoc deep OOD detection on already trained neural network image classifiers. The philosophy of the library is to favor quality over quantity and to foster easy adoption. As a result, we provide a simple, compact and easily customizable API and carefully integrate and test each proposed baseline into a coherent framework that is designed to enable their use in tensorflow **and** pytorch. You can find the documentation [here](https://deel-ai.github.io/oodeel/).
+<b>Oodeel</b> is a library that performs post-hoc deep OOD (Out-of-Distribution) detection on already trained neural network image classifiers. The philosophy of the library is to favor quality over quantity and to foster easy adoption. As a result, we provide a simple, compact and easily customizable API and carefully integrate and test each proposed baseline into a coherent framework that is designed to enable their use in tensorflow **and** pytorch. You can find the documentation [here](https://deel-ai.github.io/oodeel/).
 
 ```python
 from oodeel.methods import MLS
@@ -75,7 +183,8 @@ scores, info = mls.score(ds) # ds is a tf.data.Dataset or a torch.DataLoader
 - [Contributing](#contributing)
 - [See Also](#see-also)
 - [Acknowledgments](#acknowledgments)
-- [Creator](#creator)
+- [Creators](#creators)
+- [Citation](#citation)
 - [License](#license)
 
 # Installation
@@ -202,7 +311,11 @@ Currently, **oodeel** includes the following baselines:
 | ReAct | [ReAct: Out-of-distribution Detection With Rectified Activations](http://arxiv.org/abs/2111.12797) | NeurIPS 2021 | avail [tensorflow](docs/notebooks/tensorflow/demo_react_tf.ipynb) or  [torch](docs/notebooks/torch/demo_react_torch.ipynb) |
 | NMD | [Neural Mean Discrepancy for Efficient Out-of-Distribution Detection](https://openaccess.thecvf.com/content/CVPR2022/html/Dong_Neural_Mean_Discrepancy_for_Efficient_Out-of-Distribution_Detection_CVPR_2022_paper.html) | CVPR 2022 | planned |
 | Gram | [Detecting Out-of-Distribution Examples with Gram Matrices](https://proceedings.mlr.press/v119/sastry20a.html) | ICML 2020 | avail [tensorflow](docs/notebooks/tensorflow/demo_gram_tf.ipynb) or  [torch](docs/notebooks/torch/demo_gram_torch.ipynb) |
-
+| GEN | [GEN: Pushing the Limits of Softmax-Based Out-of-Distribution Detection](https://openaccess.thecvf.com/content/CVPR2023/html/Liu_GEN_Pushing_the_Limits_of_Softmax-Based_Out-of-Distribution_Detection_CVPR_2023_paper.html) | CVPR 2023 | avail [tensorflow](docs/notebooks/tensorflow/demo_gen_tf.ipynb) or [torch](docs/notebooks/torch/demo_gen_torch.ipynb) |
+| RMDS | [A Simple Fix to Mahalanobis Distance for Improving Near-OOD Detection](https://arxiv.org/abs/2106.09022) | preprint | avail [tensorflow](docs/notebooks/tensorflow/demo_rmds_tf.ipynb) or [torch](docs/notebooks/torch/demo_rmds_torch.ipynb) |
+| SHE | [Out-of-Distribution Detection based on In-Distribution Data Patterns Memorization with Modern Hopfield Energy](https://openreview.net/forum?id=KkazG4lgKL) | ICLR 2023 | avail [tensorflow](docs/notebooks/tensorflow/demo_she_tf.ipynb) or [torch](docs/notebooks/torch/demo_she_torch.ipynb) |
+| ASH | [Extremely Simple Activation Shaping for Out-of-Distribution Detection](http://arxiv.org/abs/2310.00227) | ICLR 2023 | avail [tensorflow](docs/notebooks/tensorflow/demo_ash_tf.ipynb) or [torch](docs/notebooks/torch/demo_ash_torch.ipynb) |
+| SCALE | [Scaling for Training Time and Post-hoc Out-of-distribution Detection Enhancement](https://arxiv.org/abs/2111.12797) | ICLR 2024 | avail [tensorflow](docs/notebooks/tensorflow/demo_scale_tf.ipynb) or [torch](docs/notebooks/torch/demo_scale_torch.ipynb) |
 
 
 
@@ -252,6 +365,19 @@ This project received funding from the French ”Investing for the Future – PI
 
 The library was created by Paul Novello to streamline DEEL research on post-hoc deep OOD methods and foster their adoption by DEEL industrial partners. He was soon joined by Yann Pequignot, Yannick Prudent, Corentin Friedrich and Matthieu Le Goff.
 
+# Citation
+
+If you use OODEEL for your research project, please consider citing:
+```
+@misc{oodeel,
+  author = {Novello, Paul and Prudent, Yannick and Friedrich, Corentin and Pequignot, Yann and Le Goff, Matthieu},
+  title = {OODEEL, a simple, compact, and hackable post-hoc deep OOD detection for already trained tensorflow or pytorch image classifiers.},
+  year = {2023},
+  publisher = {GitHub},
+  journal = {GitHub repository},
+  howpublished = {\url{https://github.com/deel-ai/oodeel}},
+}
+```
 # License
 
 The package is released under [MIT license](LICENSE).

{oodeel-0.2.0 → oodeel-0.3.0}/README.md

@@ -12,29 +12,23 @@
 <!-- Badge section -->
 <div align="center">
     <a href="#">
-        <img src="https://img.shields.io/badge/python-3.8%2B-blue">
-    </a>
+        <img src="https://img.shields.io/badge/python-3.8%2B-blue"></a>
     <a href="https://github.com/deel-ai/oodeel/actions/workflows/python-linters.yml">
-        <img alt="Flake8" src="https://github.com/deel-ai/oodeel/actions/workflows/python-linters.yml/badge.svg">
-    </a>
+        <img alt="Flake8" src="https://github.com/deel-ai/oodeel/actions/workflows/python-linters.yml/badge.svg"></a>
     <a href="https://github.com/deel-ai/oodeel/actions/workflows/python-tests-tf.yml">
-        <img alt="Tests tf" src="https://github.com/deel-ai/oodeel/actions/workflows/python-tests-tf.yml/badge.svg">
-    </a>
+        <img alt="Tests tf" src="https://github.com/deel-ai/oodeel/actions/workflows/python-tests-tf.yml/badge.svg"></a>
     <a href="https://github.com/deel-ai/oodeel/actions/workflows/python-tests-torch.yml">
-        <img alt="Tests torch" src="https://github.com/deel-ai/oodeel/actions/workflows/python-tests-torch.yml/badge.svg">
-    </a>
+        <img alt="Tests torch" src="https://github.com/deel-ai/oodeel/actions/workflows/python-tests-torch.yml/badge.svg"></a>
     <a href="https://github.com/deel-ai/oodeel/actions/workflows/python-coverage-shield.yml">
-        <img alt="Coverage" src="https://github.com/deel-ai/oodeel/raw/gh-shields/coverage.svg">
-    </a>
+        <img alt="Coverage" src="https://github.com/deel-ai/oodeel/raw/gh-shields/coverage.svg"></a>
     <a href="https://github.com/deel-ai/oodeel/blob/master/LICENSE">
-        <img alt="License MIT" src="https://img.shields.io/badge/License-MIT-efefef">
-    </a>
+        <img alt="License MIT" src="https://img.shields.io/badge/License-MIT-efefef"></a>
 </div>
 <br>
 
 <!-- Short description of your library -->
 
-<b>Oodeel</b> is a library that performs post-hoc deep OOD detection on already trained neural network image classifiers. The philosophy of the library is to favor quality over quantity and to foster easy adoption. As a result, we provide a simple, compact and easily customizable API and carefully integrate and test each proposed baseline into a coherent framework that is designed to enable their use in tensorflow **and** pytorch. You can find the documentation [here](https://deel-ai.github.io/oodeel/).
+<b>Oodeel</b> is a library that performs post-hoc deep OOD (Out-of-Distribution) detection on already trained neural network image classifiers. The philosophy of the library is to favor quality over quantity and to foster easy adoption. As a result, we provide a simple, compact and easily customizable API and carefully integrate and test each proposed baseline into a coherent framework that is designed to enable their use in tensorflow **and** pytorch. You can find the documentation [here](https://deel-ai.github.io/oodeel/).
 
 ```python
 from oodeel.methods import MLS
@@ -54,7 +48,8 @@ scores, info = mls.score(ds) # ds is a tf.data.Dataset or a torch.DataLoader
 - [Contributing](#contributing)
 - [See Also](#see-also)
 - [Acknowledgments](#acknowledgments)
-- [Creator](#creator)
+- [Creators](#creators)
+- [Citation](#citation)
 - [License](#license)
 
 # Installation
@@ -181,7 +176,11 @@ Currently, **oodeel** includes the following baselines:
 | ReAct | [ReAct: Out-of-distribution Detection With Rectified Activations](http://arxiv.org/abs/2111.12797) | NeurIPS 2021 | avail [tensorflow](docs/notebooks/tensorflow/demo_react_tf.ipynb) or  [torch](docs/notebooks/torch/demo_react_torch.ipynb) |
 | NMD | [Neural Mean Discrepancy for Efficient Out-of-Distribution Detection](https://openaccess.thecvf.com/content/CVPR2022/html/Dong_Neural_Mean_Discrepancy_for_Efficient_Out-of-Distribution_Detection_CVPR_2022_paper.html) | CVPR 2022 | planned |
 | Gram | [Detecting Out-of-Distribution Examples with Gram Matrices](https://proceedings.mlr.press/v119/sastry20a.html) | ICML 2020 | avail [tensorflow](docs/notebooks/tensorflow/demo_gram_tf.ipynb) or  [torch](docs/notebooks/torch/demo_gram_torch.ipynb) |
-
+| GEN | [GEN: Pushing the Limits of Softmax-Based Out-of-Distribution Detection](https://openaccess.thecvf.com/content/CVPR2023/html/Liu_GEN_Pushing_the_Limits_of_Softmax-Based_Out-of-Distribution_Detection_CVPR_2023_paper.html) | CVPR 2023 | avail [tensorflow](docs/notebooks/tensorflow/demo_gen_tf.ipynb) or [torch](docs/notebooks/torch/demo_gen_torch.ipynb) |
+| RMDS | [A Simple Fix to Mahalanobis Distance for Improving Near-OOD Detection](https://arxiv.org/abs/2106.09022) | preprint | avail [tensorflow](docs/notebooks/tensorflow/demo_rmds_tf.ipynb) or [torch](docs/notebooks/torch/demo_rmds_torch.ipynb) |
+| SHE | [Out-of-Distribution Detection based on In-Distribution Data Patterns Memorization with Modern Hopfield Energy](https://openreview.net/forum?id=KkazG4lgKL) | ICLR 2023 | avail [tensorflow](docs/notebooks/tensorflow/demo_she_tf.ipynb) or [torch](docs/notebooks/torch/demo_she_torch.ipynb) |
+| ASH | [Extremely Simple Activation Shaping for Out-of-Distribution Detection](http://arxiv.org/abs/2310.00227) | ICLR 2023 | avail [tensorflow](docs/notebooks/tensorflow/demo_ash_tf.ipynb) or [torch](docs/notebooks/torch/demo_ash_torch.ipynb) |
+| SCALE | [Scaling for Training Time and Post-hoc Out-of-distribution Detection Enhancement](https://arxiv.org/abs/2111.12797) | ICLR 2024 | avail [tensorflow](docs/notebooks/tensorflow/demo_scale_tf.ipynb) or [torch](docs/notebooks/torch/demo_scale_torch.ipynb) |
 
 
 
@@ -231,6 +230,19 @@ This project received funding from the French ”Investing for the Future – PI
 
 The library was created by Paul Novello to streamline DEEL research on post-hoc deep OOD methods and foster their adoption by DEEL industrial partners. He was soon joined by Yann Pequignot, Yannick Prudent, Corentin Friedrich and Matthieu Le Goff.
 
+# Citation
+
+If you use OODEEL for your research project, please consider citing:
+```
+@misc{oodeel,
+  author = {Novello, Paul and Prudent, Yannick and Friedrich, Corentin and Pequignot, Yann and Le Goff, Matthieu},
+  title = {OODEEL, a simple, compact, and hackable post-hoc deep OOD detection for already trained tensorflow or pytorch image classifiers.},
+  year = {2023},
+  publisher = {GitHub},
+  journal = {GitHub repository},
+  howpublished = {\url{https://github.com/deel-ai/oodeel}},
+}
+```
 # License
 
 The package is released under [MIT license](LICENSE).

{oodeel-0.2.0 → oodeel-0.3.0}/oodeel/__init__.py

@@ -25,4 +25,4 @@ oodeel
 -------
 """
 
-__version__ = "0.2.0"
+__version__ = "0.3.0"

{oodeel-0.2.0 → oodeel-0.3.0}/oodeel/datasets/__init__.py

@@ -20,4 +20,5 @@
 # LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 # OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 # SOFTWARE.
-from .ooddataset import OODDataset
+from .data_handler import load_data_handler
+from .deprecated.DEPRECATED_ooddataset import OODDataset

oodeel-0.3.0/oodeel/datasets/data_handler.py

@@ -0,0 +1,304 @@
+# -*- coding: utf-8 -*-
+# Copyright IRT Antoine de Saint Exupéry et Université Paul Sabatier Toulouse III - All
+# rights reserved. DEEL is a research program operated by IVADO, IRT Saint Exupéry,
+# CRIAQ and ANITI - https://www.deel.ai/
+#
+# Permission is hereby granted, free of charge, to any person obtaining a copy
+# of this software and associated documentation files (the "Software"), to deal
+# in the Software without restriction, including without limitation the rights
+# to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+# copies of the Software, and to permit persons to whom the Software is
+# furnished to do so, subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in all
+# copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+# FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+# SOFTWARE.
+import importlib.util
+from abc import ABC
+from abc import abstractmethod
+
+import numpy as np
+
+from ..types import Callable
+from ..types import DatasetType
+from ..types import ItemType
+from ..types import Optional
+from ..types import TensorType
+from ..types import Tuple
+from ..types import Union
+
+
+def get_backend():
+    """Detects whether TensorFlow or PyTorch is available and returns
+    the preferred backend."""
+    available_backends = []
+    if importlib.util.find_spec("tensorflow"):
+        available_backends.append("tensorflow")
+    if importlib.util.find_spec("torch"):
+        available_backends.append("torch")
+
+    if len(available_backends) == 1:
+        return available_backends[0]
+    elif len(available_backends) == 0:
+        raise ImportError("Neither TensorFlow nor PyTorch is installed.")
+    else:
+        raise ImportError(
+            "Both TensorFlow and PyTorch are installed. Please specify the backend."
+        )
+
+
+def load_data_handler(backend: str = None):
+    if backend is None:
+        backend = get_backend()
+
+    if backend == "tensorflow":
+        from .tf_data_handler import TFDataHandler
+
+        return TFDataHandler()
+
+    elif backend == "torch":
+        from .torch_data_handler import TorchDataHandler
+
+        return TorchDataHandler()
+
+
+class DataHandler(ABC):
+    """
+    Class to manage Datasets. The aim is to provide a simple interface
+    for working with datasets (torch, tensorflow or other...) and manage them without
+    having to use library-specific syntax.
+    """
+
+    def __init__(self):
+        self.backend = None
+        self.channel_order = None
+
+    def split_by_class(
+        self,
+        dataset: DatasetType,
+        in_labels: Optional[Union[np.ndarray, list]] = None,
+        out_labels: Optional[Union[np.ndarray, list]] = None,
+    ) -> Optional[Tuple[DatasetType]]:
+        """Filter the dataset by assigning ood labels depending on labels
+        value (typically, class id).
+
+        Args:
+            in_labels (Optional[Union[np.ndarray, list]], optional): set of labels
+                to be considered as in-distribution. Defaults to None.
+            out_labels (Optional[Union[np.ndarray, list]], optional): set of labels
+                to be considered as out-of-distribution. Defaults to None.
+
+        Returns:
+            Optional[Tuple[OODDataset]]: Tuple of in-distribution and
+                out-of-distribution OODDatasets
+        """
+        # Make sure the dataset has labels
+        assert (in_labels is not None) or (
+            out_labels is not None
+        ), "specify labels to filter with"
+        assert self.get_item_length(dataset) >= 2, "the dataset has no labels"
+
+        # Filter the dataset depending on in_labels and out_labels given
+        if (out_labels is not None) and (in_labels is not None):
+            in_data = self.filter_by_value(dataset, "label", in_labels)
+            out_data = self.filter_by_value(dataset, "label", out_labels)
+
+        if out_labels is None:
+            in_data = self.filter_by_value(dataset, "label", in_labels)
+            out_data = self.filter_by_value(dataset, "label", in_labels, excluded=True)
+
+        elif in_labels is None:
+            in_data = self.filter_by_value(dataset, "label", out_labels, excluded=True)
+            out_data = self.filter_by_value(dataset, "label", out_labels)
+
+        # Return the filtered OODDatasets
+        return in_data, out_data
+
+    @classmethod
+    @abstractmethod
+    def prepare(
+        cls,
+        dataset: DatasetType,
+        batch_size: int,
+        preprocess_fn: Optional[Callable] = None,
+        augment_fn: Optional[Callable] = None,
+        columns: Optional[list] = None,
+        shuffle: bool = False,
+        dict_based_fns: bool = True,
+        return_tuple: bool = True,
+        **kwargs_prepare,
+    ) -> DatasetType:
+        """Prepare dataset for scoring or training
+
+        Args:
+            batch_size (int): Batch size
+            preprocess_fn (Callable, optional): Preprocessing function to apply to
+                the dataset. Defaults to None.
+            augment_fn (Callable, optional): Augment function to be used (when the
+                returned dataset is to be used for training). Defaults to None.
+            columns (list, optional): List of columns
+                that will be returned. Keep all columns if None. Defaults to None.
+            shuffle (bool, optional): To shuffle the returned dataset or not.
+                Defaults to False.
+            dict_based_fns (bool): Whether to use preprocess and DA functions as dict
+                based (if True) or as tuple based (if False). Defaults to True.
+            return_tuple (bool, optional): Whether to return each dataset item
+                as a tuple. Defaults to True.
+            kwargs_prepare (dict): Additional parameters to be passed to the
+                data_handler for backend specific preparation.
+
+
+        Returns:
+            DatasetType: prepared dataset
+        """
+        raise NotImplementedError()
+
+    @staticmethod
+    @abstractmethod
+    def load_dataset_from_arrays(
+        dataset_id: ItemType, columns: Optional[list] = None
+    ) -> DatasetType:
+        """Load a DatasetType from a np.ndarray / Tensor
+
+        Args:
+            dataset_id (ItemType): numpy array(s) to load.
+            columns (list, optional): Column names to assign. If None,
+                assigned as "input_i" for i-th column. Defaults to None.
+
+        Returns:
+            DatasetType
+        """
+        raise NotImplementedError()
+
+    @staticmethod
+    @abstractmethod
+    def load_custom_dataset(
+        dataset_id: DatasetType, columns: Optional[list] = None
+    ) -> DatasetType:
+        """Load a custom dataset by ensuring it is properly formatted.
+
+        Args:
+            dataset_id (DatasetType): dataset
+            columns (list, optional): Column names to use for elements if dataset_id is
+                tuple based. If None, assigned as "input_i"
+                for i-th column. Defaults to None.
+
+        Returns:
+            A properly formatted dataset.
+        """
+        raise NotImplementedError()
+
+    @staticmethod
+    @abstractmethod
+    def map_ds(dataset: DatasetType, map_fn: Callable) -> DatasetType:
+        """Map a function to a Dataset
+
+        Args:
+            dataset (DatasetType): Dataset to map the function to
+            map_fn (Callable): Function to map
+
+        Returns:
+            DatasetType: Mapped dataset
+        """
+        raise NotImplementedError()
+
+    @staticmethod
+    @abstractmethod
+    def filter_by_value(
+        dataset: DatasetType,
+        column_name: str,
+        values: list,
+        excluded: bool = False,
+    ) -> DatasetType:
+        """Filter the dataset by checking the value of a column is in `values`
+
+        Args:
+            dataset (Dataset): Dataset to filter
+            column_name (str): Column to filter the dataset with
+            values (list): Column values to keep (if excluded is False)
+                or to exclude
+            excluded (bool, optional): To keep (False) or exclude (True) the samples
+                with column value included in Values. Defaults to False.
+
+        Returns:
+            DatasetType: Filtered dataset
+        """
+        raise NotImplementedError()
+
+    @staticmethod
+    @abstractmethod
+    def get_item_length(dataset: DatasetType) -> int:
+        """Number of elements in a dataset item
+
+        Args:
+            dataset (DatasetType): Dataset
+
+        Returns:
+            int: Item length
+        """
+        raise NotImplementedError()
+
+    @staticmethod
+    @abstractmethod
+    def get_dataset_length(dataset: DatasetType) -> int:
+        """Number of items in a dataset
+
+        Args:
+            dataset (DatasetType): Dataset
+
+        Returns:
+            int: Dataset length
+        """
+        raise NotImplementedError()
+
+    @staticmethod
+    @abstractmethod
+    def get_column_elements_shape(
+        dataset: DatasetType, column_name: Union[str, int]
+    ) -> tuple:
+        """Get the shape of the elements of a column of dataset identified by
+        column_name
+
+        Args:
+            dataset (Dataset): a Dataset
+            column_name (Union[str, int]): The column name to get
+                the element shape from.
+
+        Returns:
+            tuple: the shape of an element from column_name
+        """
+        raise NotImplementedError()
+
+    @staticmethod
+    @abstractmethod
+    def get_input_from_dataset_item(elem: ItemType) -> TensorType:
+        """Get the tensor that is to be feed as input to a model from a dataset element.
+
+        Args:
+            elem (ItemType): dataset element to extract input from
+
+        Returns:
+            TensorType: Input tensor
+        """
+        raise NotImplementedError()
+
+    @staticmethod
+    @abstractmethod
+    def get_label_from_dataset_item(item: ItemType):
+        """Retrieve label tensor from item as a tuple/list. Label must be at index 1
+        in the item tuple. If one-hot encoded, labels are converted to single value.
+
+        Args:
+            elem (ItemType): dataset element to extract label from
+
+        Returns:
+            Any: Label tensor
+        """
+        raise NotImplementedError()

oodeel-0.2.0/oodeel/datasets/data_handler.py → oodeel-0.3.0/oodeel/datasets/deprecated/DEPRECATED_data_handler.py

@@ -25,12 +25,12 @@ from abc import abstractmethod
 
 import numpy as np
 
-from ..types import Callable
-from ..types import DatasetType
-from ..types import ItemType
-from ..types import Optional
-from ..types import Tuple
-from ..types import Union
+from ...types import Callable
+from ...types import DatasetType
+from ...types import ItemType
+from ...types import Optional
+from ...types import Tuple
+from ...types import Union
 
 
 class DataHandler(ABC):