chore: merged dev branch to ConvNeXt3D PR

frankkramer-lab · Feb 21, 2024 · 729a275 · 729a275
2 parents e0d6779 + 439683a
commit 729a275
Show file tree

Hide file tree

Showing 26 changed files with 117 additions and 161 deletions.
diff --git a/.github/workflows/build-package.yml b/.github/workflows/build-package.yml
@@ -17,7 +17,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        python-version: ['3.8', '3.10']
+        python-version: ['3.9', '3.10']
     steps:
     - uses: actions/checkout@v2
     - name: Set up Python ${{ matrix.python-version }}
@@ -39,7 +39,7 @@ jobs:
     strategy:
       fail-fast: false
       matrix:
-        python-version: ['3.8', '3.10']
+        python-version: ['3.9', '3.10']
     steps:
     - uses: actions/checkout@v2
     - name: Set up Python ${{ matrix.python-version }}

diff --git a/.github/workflows/code-quality.yml b/.github/workflows/code-quality.yml
@@ -15,10 +15,10 @@ jobs:
     runs-on: ubuntu-latest
     steps:
     - uses: actions/checkout@v2
-    - name: Set up Python 3.8
+    - name: Set up Python 3.10
       uses: actions/setup-python@v2
       with:
-        python-version: 3.8
+        python-version: '3.10'
     - name: Install dependencies
       run: |
         python -m pip install --upgrade pip

diff --git a/.github/workflows/release.yml b/.github/workflows/release.yml
@@ -96,7 +96,7 @@ jobs:
     - uses: actions/checkout@v2
     - uses: actions/setup-python@v2
       with:
-        python-version: 3.8
+        python-version: '3.10'
     - name: Install dependencies
       run: |
         python -m pip install --upgrade pip

diff --git a/.github/workflows/website.yml b/.github/workflows/website.yml
@@ -10,7 +10,7 @@ jobs:
       - uses: actions/checkout@v2
       - uses: actions/setup-python@v2
         with:
-          python-version: 3.8
+          python-version: '3.10'
       - run: pip install \
           black \
           mkdocs-material \

diff --git a/aucmedi/ensemble/bagging.py b/aucmedi/ensemble/bagging.py
@@ -159,6 +159,8 @@ def train(self, training_generator, epochs=20, iterations=None,
                 data = (train_x, train_y, None, test_x, test_y, None)
             else : data = fold
 
+            # Create model specific callback list
+            callbacks_model = callbacks.copy()
             # Extend Callback list
             cb_mc = ModelCheckpoint(os.path.join(self.cache_dir.name,
                                                  "cv_" + str(i) + \
@@ -169,7 +171,7 @@ def train(self, training_generator, epochs=20, iterations=None,
                                                  "cv_" + str(i) + \
                                                  ".logs.csv"),
                               separator=',', append=True)
-            callbacks.extend([cb_mc, cb_cl])
+            callbacks_model.extend([cb_mc, cb_cl])
 
             # Gather NeuralNetwork parameters
             model_paras = {
@@ -210,7 +212,7 @@ def train(self, training_generator, epochs=20, iterations=None,
             # Gather training parameters
             parameters_training = {"epochs": epochs,
                                    "iterations": iterations,
-                                   "callbacks": callbacks,
+                                   "callbacks": callbacks_model,
                                    "class_weights": class_weights,
                                    "transfer_learning": transfer_learning
             }
@@ -298,12 +300,14 @@ def predict(self, prediction_generator, aggregate="mean",
                          "kwargs": temp_dg.kwargs
         }
 
+        # Identify path to model directory
+        if isinstance(self.cache_dir, tempfile.TemporaryDirectory):
+            path_model_dir = self.cache_dir.name
+        else : path_model_dir = self.cache_dir
+
         # Sequentially iterate over all fold models
         for i in range(self.k_fold):
             # Identify path to fitted model
-            if isinstance(self.cache_dir, tempfile.TemporaryDirectory):
-                path_model_dir = self.cache_dir.name
-            else : path_model_dir = self.cache_dir
             path_model = os.path.join(path_model_dir,
                                       "cv_" + str(i) + ".model.hdf5")
 

diff --git a/aucmedi/ensemble/composite.py b/aucmedi/ensemble/composite.py
@@ -207,14 +207,6 @@ def train(self, training_generator, epochs=20, iterations=None,
         cv_sampling = sampling_kfold(x, y, m, n_splits=self.k_fold,
                                      stratified=True, iterative=True)
 
-        # Gather training parameters
-        parameters_training = {"epochs": epochs,
-                               "iterations": iterations,
-                               "callbacks": callbacks,
-                               "class_weights": class_weights,
-                               "transfer_learning": transfer_learning
-        }
-
         # Sequentially iterate over model list
         for i in range(len(self.model_list)):
             # Pack data into a tuple
@@ -224,6 +216,8 @@ def train(self, training_generator, epochs=20, iterations=None,
                 data = (train_x, train_y, None, test_x, test_y, None)
             else : data = fold
 
+            # Create model specific callback list
+            callbacks_model = callbacks.copy()
             # Extend Callback list
             path_model = os.path.join(self.cache_dir.name,
                                       "cv_" + str(i) + ".model.hdf5")
@@ -234,7 +228,7 @@ def train(self, training_generator, epochs=20, iterations=None,
                                                  "cv_" + str(i) + \
                                                  ".logs.csv"),
                               separator=',', append=True)
-            callbacks.extend([cb_mc, cb_cl])
+            callbacks_model.extend([cb_mc, cb_cl])
 
             # Gather NeuralNetwork parameters
             model_paras = {
@@ -272,6 +266,14 @@ def train(self, training_generator, epochs=20, iterations=None,
                              "kwargs": temp_dg.kwargs
             }
 
+            # Gather training parameters
+            parameters_training = {"epochs": epochs,
+                                "iterations": iterations,
+                                "callbacks": callbacks_model,
+                                "class_weights": class_weights,
+                                "transfer_learning": transfer_learning
+            }
+
             # Start training process
             process_queue = mp.Queue()
             process_train = mp.Process(target=__training_process__,
@@ -335,7 +337,7 @@ def train_metalearner(self, training_generator):
 
         # Sequentially iterate over model list
         for i in range(len(self.model_list)):
-            # Extend Callback list
+            # Load current model
             path_model = os.path.join(path_model_dir,
                                       "cv_" + str(i) + ".model.hdf5")
 

diff --git a/aucmedi/ensemble/stacking.py b/aucmedi/ensemble/stacking.py
@@ -202,16 +202,10 @@ def train(self, training_generator, epochs=20, iterations=None,
             data_train = (*ps_sampling[0], None)
             data_val = (*ps_sampling[1], None)
 
-        # Gather training parameters
-        parameters_training = {"epochs": epochs,
-                               "iterations": iterations,
-                               "callbacks": callbacks,
-                               "class_weights": class_weights,
-                               "transfer_learning": transfer_learning
-        }
-
         # Sequentially iterate over model list
         for i in range(len(self.model_list)):
+            # Create model specific callback list
+            callbacks_model = callbacks.copy()
             # Extend Callback list
             path_model = os.path.join(self.cache_dir.name,
                                       "nn_" + str(i) + ".model.hdf5")
@@ -222,7 +216,7 @@ def train(self, training_generator, epochs=20, iterations=None,
                                                  "nn_" + str(i) + \
                                                  ".logs.csv"),
                               separator=',', append=True)
-            callbacks.extend([cb_mc, cb_cl])
+            callbacks_model.extend([cb_mc, cb_cl])
 
             # Gather NeuralNetwork parameters
             model_paras = {
@@ -260,6 +254,14 @@ def train(self, training_generator, epochs=20, iterations=None,
                              "kwargs": temp_dg.kwargs
             }
 
+            # Gather training parameters
+            parameters_training = {"epochs": epochs,
+                                "iterations": iterations,
+                                "callbacks": callbacks_model,
+                                "class_weights": class_weights,
+                                "transfer_learning": transfer_learning
+            }
+
             # Start training process
             process_queue = mp.Queue()
             process_train = mp.Process(target=__training_process__,
@@ -323,7 +325,7 @@ def train_metalearner(self, training_generator):
 
         # Sequentially iterate over model list
         for i in range(len(self.model_list)):
-            # Extend Callback list
+            #  Load current model
             path_model = os.path.join(path_model_dir,
                                       "nn_" + str(i) + ".model.hdf5")
 

diff --git a/aucmedi/evaluation/comparison.py b/aucmedi/evaluation/comparison.py
@@ -147,7 +147,7 @@ def evaluate_comparison(pred_list,
 
         # Optional: Macro average classes
         if macro_average_classes:
-            metrics_avg = metrics.groupby(["metric", "model"]).mean()
+            metrics_avg = metrics.groupby(["metric", "model"])[["score"]].mean()
             metrics = metrics_avg.reset_index()
 
         # Append to dataframe list

diff --git a/aucmedi/evaluation/performance.py b/aucmedi/evaluation/performance.py
@@ -236,7 +236,7 @@ def evalby_barplot(metrics, out_path, class_names, show=False, suffix=None):
 #-----------------------------------------------------#
 def evalby_rocplot(fpr_list, tpr_list, out_path, class_names, show=False, suffix=None):
     # Initialize result dataframe
-    df_roc = pd.DataFrame(data=[fpr_list, tpr_list], dtype=np.float64)
+    df_roc = pd.DataFrame(data=[fpr_list, tpr_list], dtype=object)
     # Preprocess dataframe
     df_roc = df_roc.transpose()
     df_roc = df_roc.apply(pd.Series.explode)

diff --git a/aucmedi/neural_network/architectures/image/__init__.py b/aucmedi/neural_network/architectures/image/__init__.py
@@ -49,9 +49,6 @@
 from aucmedi.neural_network.architectures.image.resnet50v2 import ResNet50V2
 from aucmedi.neural_network.architectures.image.resnet101v2 import ResNet101V2
 from aucmedi.neural_network.architectures.image.resnet152v2 import ResNet152V2
-# ResNeXt
-from aucmedi.neural_network.architectures.image.resnext50 import ResNeXt50
-from aucmedi.neural_network.architectures.image.resnext101 import ResNeXt101
 # MobileNet
 from aucmedi.neural_network.architectures.image.mobilenet import MobileNet
 from aucmedi.neural_network.architectures.image.mobilenetv2 import MobileNetV2
@@ -86,8 +83,6 @@
     "ResNet50V2": ResNet50V2,
     "ResNet101V2": ResNet101V2,
     "ResNet152V2": ResNet152V2,
-    "ResNeXt50": ResNeXt50,
-    "ResNeXt101": ResNeXt101,
     "DenseNet121": DenseNet121,
     "DenseNet169": DenseNet169,
     "DenseNet201": DenseNet201,
@@ -153,8 +148,8 @@
 
         For example:
         ```python
-        # for the image architecture "ResNeXt101"
-        architecture="2D.ResNeXt101"
+        # for the image architecture "ResNet101"
+        architecture="2D.ResNet101"
         ```
 
     Architectures are based on the abstract base class [aucmedi.neural_network.architectures.arch_base.Architecture_Base][].
@@ -175,8 +170,6 @@
     "ResNet50V2": "tf",
     "ResNet101V2": "tf",
     "ResNet152V2": "tf",
-    "ResNeXt50": "torch",
-    "ResNeXt101": "torch",
     "DenseNet121": "torch",
     "DenseNet169": "torch",
     "DenseNet201": "torch",
@@ -237,8 +230,8 @@
 
         For example:
         ```python
-        # for the image architecture "ResNeXt101"
+        # for the image architecture "ResNet101"
         from aucmedi.neural_network.architectures import supported_standardize_mode
-        sf_norm = supported_standardize_mode["2D.ResNeXt101"]
+        sf_norm = supported_standardize_mode["2D.ResNet101"]
         ```
 """
diff --git a/aucmedi/neural_network/architectures/image/resnext101.py b/aucmedi/neural_network/architectures/image/resnext101.py
@@ -1,6 +1,6 @@
 #==============================================================================#
 #  Author:       Dominik Müller                                                #
-#  Copyright:    2023 IT-Infrastructure for Translational Medical Research,    #
+#  Copyright:    2022 IT-Infrastructure for Translational Medical Research,    #
 #                University of Augsburg                                        #
 #                                                                              #
 #  This program is free software: you can redistribute it and/or modify        #

diff --git a/aucmedi/neural_network/architectures/image/resnext50.py b/aucmedi/neural_network/architectures/image/resnext50.py
@@ -1,6 +1,6 @@
 #==============================================================================#
 #  Author:       Dominik Müller                                                #
-#  Copyright:    2023 IT-Infrastructure for Translational Medical Research,    #
+#  Copyright:    2022 IT-Infrastructure for Translational Medical Research,    #
 #                University of Augsburg                                        #
 #                                                                              #
 #  This program is free software: you can redistribute it and/or modify        #

diff --git a/aucmedi/utils/class_weights.py b/aucmedi/utils/class_weights.py
@@ -100,7 +100,7 @@ def compute_multilabel_weights(ohe_array, method="balanced"):
     class_weights = np.empty([n_classes])
     # Compute weight for each class individually
     for i in range(0, n_classes):
-        weight = compute_class_weight(class_weight=method, classes=[0,1],
+        weight = compute_class_weight(class_weight=method, classes=np.array([0,1]),
                                       y=ohe_array[:, i])
         class_weights[i] = weight[1]
     # Return resulting class weight list

diff --git a/docs/examples/tutorials.md b/docs/examples/tutorials.md
@@ -8,7 +8,7 @@ For this reason, rather than creating one huge “demonstration Notebook” seve
 
 | Topic | Link  |
 |:------|:-----:|
-| Three Pillars of AUCMEDI | [tutorial01.ipynb](https://github.com/frankkramer-lab/aucmedi/blob/master/examples/tutorials/Augmentation.ipynb) |
+| Three Pillars of AUCMEDI | [tutorial01.ipynb](https://github.com/frankkramer-lab/aucmedi/blob/master/examples/tutorials/ThreePillars.ipynb) |
 | Data Loading from CSV | [tutorial02.ipynb](https://github.com/frankkramer-lab/aucmedi/blob/master/examples/tutorials/dataLoadingFromCSV.ipynb) |
 | Data Loading from Directories | [tutorial03.ipynb](https://github.com/frankkramer-lab/aucmedi/blob/master/examples/tutorials/dataLoadingFromDirectory.ipynb) |
 | Architecture Selection | [tutorial04.ipynb](https://github.com/frankkramer-lab/aucmedi/blob/master/examples/tutorials/CustomArchitecture.ipynb) |

diff --git a/docs/getstarted/setup.md b/docs/getstarted/setup.md
@@ -3,7 +3,7 @@
 AUCMEDI is tested and supported on the following systems:
 
 - 64-bit Ubuntu 16.04 or later  
-- Python 3.8 - 3.10  
+- Python 3.9+ 
 
 AUCMEDI is heavily based on TensorFlow 2: an approachable, highly-productive interface for solving machine learning problems, with a focus on modern deep learning. It provides essential abstractions and building blocks for developing and shipping machine learning solutions with high iteration velocity.
 

diff --git a/examples/tutorials/Augmentation.ipynb b/examples/tutorials/Augmentation.ipynb
diff --git a/examples/tutorials/DataAndResultExploration.ipynb b/examples/tutorials/DataAndResultExploration.ipynb
diff --git a/examples/tutorials/EarlyStoppingAndTransferlearning.ipynb b/examples/tutorials/EarlyStoppingAndTransferlearning.ipynb
diff --git a/examples/tutorials/Metadata.ipynb b/examples/tutorials/Metadata.ipynb
@@ -405,7 +405,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.8.10"
+   "version": "3.8.0"
   }
  },
  "nbformat": 4,

diff --git a/examples/tutorials/ThreePillars.ipynb b/examples/tutorials/ThreePillars.ipynb
diff --git a/examples/tutorials/Transferlearning.ipynb b/examples/tutorials/Transferlearning.ipynb
diff --git a/examples/tutorials/XAI.ipynb b/examples/tutorials/XAI.ipynb
diff --git a/examples/tutorials/dataLoadingFromDirectory.ipynb b/examples/tutorials/dataLoadingFromDirectory.ipynb
diff --git a/requirements.txt b/requirements.txt
@@ -1,19 +1,18 @@
-tensorflow==2.11.0
-tensorflow-addons==0.18.0
-keras-applications==1.0.8
-numpy==1.23.0
+tensorflow==2.14.0
+numpy==1.23.5
 pillow==9.3.0
 albumentations==1.3.0
-pandas==1.5.0
-scikit-learn==1.1.0
-scikit-image==0.19.2
+pandas==1.5.2
+scikit-learn==1.3.0
+scikit-image==0.21.0
 lime==0.2.0.1
 pooch==1.6.0
-classification-models-3D==1.0.7
-vit-keras==0.1.0
+classification-models-3D==1.0.10
+vit-keras==0.1.2
+tensorflow-addons==0.21.0
+Keras-Applications==1.0.8
 SimpleITK==2.2.0
-batchgenerators==0.24
+batchgenerators==0.25
 volumentations-aucmedi==1.0.1
-plotnine==0.10.0
+plotnine==0.12.4
 pathos==0.3.0
-matplotlib==3.5.1
diff --git a/setup.py b/setup.py
@@ -24,28 +24,26 @@
     entry_points={
         'console_scripts': ['aucmedi = aucmedi.automl.main:main'],
     },
-    python_requires='>=3.8',
-    install_requires=['tensorflow>=2.10.0',
-                      'tensorflow-addons>=0.18.0',
-                      'keras-applications>=1.0.8',
-                      'numpy>=1.23.0',
+    python_requires='>=3.9',
+    install_requires=['tensorflow>=2.14.0',
+                      'numpy>=1.23.5',
                       'pillow>=9.3.0',
                       'albumentations>=1.3.0',
-                      'pandas>=1.5.0',
-                      'scikit-learn>=1.1.0',
-                      'scikit-image>=0.19.1',
+                      'pandas>=1.5.2',
+                      'scikit-learn>=1.3.0',
+                      'scikit-image>=0.21.0',
                       'lime>=0.2.0.1',
                       'pooch>=1.6.0',
-                      'classification-models-3D>=1.0.7',
-                      'vit-keras>=0.1.0',
+                      'classification-models-3D>=1.0.10',
+                      'vit-keras>=0.1.2',
+                      'tensorflow-addons>=0.21.0',
+                      'Keras-Applications==1.0.8',
                       'SimpleITK>=2.2.0',
-                      'batchgenerators>=0.24',
+                      'batchgenerators>=0.25',
                       'volumentations-aucmedi>=1.0.1',
-                      'plotnine>=0.10.0',
-                      'pathos>=0.3.0',
-                      'matplotlib>=3.5.0,<3.5.3'],
+                      'plotnine==0.12.4',
+                      'pathos>=0.3.0'],
     classifiers=["Programming Language :: Python :: 3",
-                 "Programming Language :: Python :: 3.8",
                  "Programming Language :: Python :: 3.10",
                  "License :: OSI Approved :: GNU General Public License v3 (GPLv3)",
                  "Operating System :: OS Independent",