Update requirements

.github/workflows/test_and_deploy.yml

@@ -0,0 +1,58 @@
+name: tests
+
+on:
+  push:
+    branches:
+      - main
+    tags:
+      - "v*" # Push events to matching v*, i.e. v1.0, v20.15.10
+  pull_request:
+    branches:
+      - main
+  workflow_dispatch:
+
+jobs:
+  test:
+    name: ${{ matrix.platform }} py${{ matrix.python-version }}
+    runs-on: ${{ matrix.platform }}
+    strategy:
+      matrix:
+        platform: [ubuntu-latest, windows-latest, macos-latest]
+        python-version: [3.7, 3.8, 3.9, '3.10']
+
+    steps:
+      - uses: actions/checkout@v3
+
+      - name: Set up Python ${{ matrix.python-version }}
+        uses: actions/setup-python@v3
+        with:
+          python-version: ${{ matrix.python-version }}
+
+      - name: Install dependencies
+        run: |
+          python -m pip install --upgrade pip
+          pip install setuptools tox tox-gh-actions
+      - name: Test with tox
+        run: tox
+  deploy:
+    needs: [test]
+    runs-on: ubuntu-latest
+    if: contains(github.ref, 'tags')
+    steps:
+      - uses: actions/checkout@v3
+      - name: Set up Python
+        uses: actions/setup-python@v3
+        with:
+          python-version: "3.x"
+      - name: Install dependencies
+        run: |
+          python -m pip install --upgrade pip
+          pip install -U setuptools setuptools_scm wheel twine build
+      - name: Build and publish
+        env:
+          TWINE_USERNAME: __token__
+          TWINE_PASSWORD: ${{ secrets.TWINE_API_KEY }}
+        run: |
+          git tag
+          python -m build .
+          twine upload dist/*

README.md

@@ -24,23 +24,31 @@ OpenReview: [https://openreview.net/forum?id=IZfQYb4lHVq](https://openreview.net
 
 ## Installation
 This implementation requires [Tensorflow](https://www.tensorflow.org/install/).
-We have tested Noise2Void using Python 3.7 and tensorflow-gpu 2.4.1.
+We have tested Noise2Void using Python 3.9 and TensorFlow 2.7 and 2.10.
 
-Note: If you want to use TensorFlow 1.15 you have to install N2V v0.2.1. N2V v0.3.0 supports TensorFlow 2 only.
+Note: If you want to use TensorFlow 1.15 you have to install N2V v0.2.1. N2V v0.3.* supports TensorFlow 2 only.
 
 #### If you start from scratch...
 We recommend using [miniconda](https://docs.conda.io/en/latest/miniconda.html).
 If you do not yet have a strong opinion, just use it too!
 
-After installing Miniconda, the following lines might are likely the easiest way to get Tensorflow and CuDNN installed on your machine (_Note:_ Macs are not supported, and if you sit on a Windows machine all this might also require some modifications.):
+After installing Miniconda, create a conda environment:
 
 ```
-conda create -n n2v python=3.8 cudatoolkit=11.0 cudnn=8.0 -c conda-forge
+conda create -n 'n2v' python=3.9
 conda activate n2v
-pip install tensorflow==2.4
 ```
 
-Once this is done (or you had tensorflow et al. installed already), you can install N2V with one of the following two options:
+#### Install TensorFlow
+
+The best way to install is to follow the [Tensorflow guidelines](https://www.tensorflow.org/install/pip). 
+
+Note that running the following commands in your environment will allow you to use the GPU without having to each 
+time run an `export` command (refer to the [Tensorflow step by step](https://www.tensorflow.org/install/pip#linux_1)):
+```bash
+mkdir -p $CONDA_PREFIX/etc/conda/activate.d
+echo 'export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$CONDA_PREFIX/lib/' > $CONDA_PREFIX/etc/conda/activate.d/env_vars.sh
+```
 
 #### Option 1: PIP (current stable release)
 ```

n2v/internals/N2V_DataWrapper.py

@@ -45,12 +45,12 @@ def __init__(self, X, Y, batch_size, length, perc_pix=0.198, shape=(64, 64),
 
         if self.dims == 2:
             self.patch_sampler = self.__subpatch_sampling2D__
-            self.box_size = np.round(np.sqrt(100/perc_pix)).astype(np.int)
+            self.box_size = np.round(np.sqrt(100/perc_pix)).astype(np.int32)
             self.get_stratified_coords = self.__get_stratified_coords2D__
             self.rand_float = self.__rand_float_coords2D__(self.box_size)
         elif self.dims == 3:
             self.patch_sampler = self.__subpatch_sampling3D__
-            self.box_size = np.round(np.sqrt(100 / perc_pix)).astype(np.int)
+            self.box_size = np.round(np.sqrt(100 / perc_pix)).astype(np.int32)
             self.get_stratified_coords = self.__get_stratified_coords3D__
             self.rand_float = self.__rand_float_coords3D__(self.box_size)
         else:
@@ -100,7 +100,7 @@ def apply_structN2Vmask(self, patch, coords, dims, mask):
         each point in coords corresponds to the center of the mask.
         then for point in the mask with value=1 we assign a random value
         """
-        coords = np.array(coords).astype(np.int)
+        coords = np.array(coords).astype(np.int32)
         ndim = mask.ndim
         center = np.array(mask.shape)//2
         ## leave the center value alone

n2v/models/n2v_standard.py

@@ -28,7 +28,7 @@
     tta_forward, tta_backward
 from ..nets.unet import build_single_unet_per_channel, build_unet
 
-from tifffile import imsave
+from tifffile import imwrite
 from csbdeep.utils.six import tempfile
 import shutil
 
@@ -300,7 +300,7 @@ def prepare_for_training(self, optimizer=None, **kwargs):
         """
         if optimizer is None:
             from tensorflow.keras.optimizers import Adam
-            optimizer = Adam(lr=self.config.train_learning_rate)
+            optimizer = Adam(learning_rate=self.config.train_learning_rate)
         self.callbacks = self.prepare_model(self.keras_model, optimizer, self.config.train_loss, **kwargs)
 
         if self.basedir is not None:
@@ -486,8 +486,8 @@ def export_TF(self, name, description, authors, test_img, axes, patch_shape, fna
 
         input_file = self.logdir / 'testinput.tif'
         output_file = self.logdir / 'testoutput.tif'
-        imsave(input_file, test_img)
-        imsave(output_file, test_output)
+        imwrite(input_file, test_img)
+        imwrite(output_file, test_output)
 
         with ZipFile(fname, 'a') as myzip:
             myzip.write(yml_file, arcname=os.path.basename(yml_file))

n2v/nets/blurpool.py

@@ -60,3 +60,10 @@ def compute_output_shape(self, input_shape):
             int(np.ceil(input_shape[2] / 2)),
             input_shape[3],
         )
+
+    def get_config(self):
+        config = super().get_config()
+        config.update({
+            "pool": self.pool
+        })
+        return config

n2v/nets/unet.py

@@ -113,8 +113,9 @@ def build_unet(input_shape,
 
     final = conv(num_channels, (1,)*n_dim, activation='linear')(unet)
     if residual:
-        if not (num_channels == 1 if backend_channels_last() else num_channels
-                                                       == 1):
+        if not (num_channels == 1):
+        #if not (num_channels == 1 if backend_channels_last() else num_channels
+        #                                              == 1):
             raise ValueError("number of input and output channels must be the same for a residual net.")
         final = Add()([final, input])
     final = Activation(activation=last_activation)(final)

n2v/utils/n2v_utils.py

@@ -162,11 +162,11 @@ def identity(patch, coords, dims, structN2Vmask=None):
 def manipulate_val_data(X_val, Y_val, perc_pix=0.198, shape=(64, 64), value_manipulation=pm_uniform_withCP(5)):
     dims = len(shape)
     if dims == 2:
-        box_size = np.round(np.sqrt(100 / perc_pix)).astype(np.int)
+        box_size = np.round(np.sqrt(100 / perc_pix)).astype(np.int32)
         get_stratified_coords = dw.__get_stratified_coords2D__
         rand_float = dw.__rand_float_coords2D__(box_size)
     elif dims == 3:
-        box_size = np.round(np.sqrt(100 / perc_pix)).astype(np.int)
+        box_size = np.round(np.sqrt(100 / perc_pix)).astype(np.int32)
         get_stratified_coords = dw.__get_stratified_coords3D__
         rand_float = dw.__rand_float_coords3D__(box_size)
 

pyproject.toml

@@ -0,0 +1,3 @@
+[build-system]
+requires = ["setuptools", "wheel"]
+build-backend = "setuptools.build_meta"

setup.py

@@ -4,13 +4,12 @@
 
 _dir = path.abspath(path.dirname(__file__))
 
-with open(path.join(_dir,'n2v','version.py')) as f:
+with open(path.join(_dir, 'n2v', 'version.py')) as f:
     exec(f.read())
 
-with open(path.join(_dir,'README.md')) as f:
+with open(path.join(_dir, 'README.md')) as f:
     long_description = f.read()
 
-
 setup(name='n2v',
       version=__version__,
       description='Noise2Void allows the training of a denoising CNN from individual noisy images. This implementation'
@@ -33,27 +32,30 @@
           'Topic :: Scientific/Engineering',
           'License :: OSI Approved :: BSD License',
 
-          'Programming Language :: Python :: 3.6',
           'Programming Language :: Python :: 3.7',
+          'Programming Language :: Python :: 3.8',
+          'Programming Language :: Python :: 3.9',
+          'Programming Language :: Python :: 3.10',
       ],
 
+
       scripts=[
           'scripts/trainN2V.py',
           'scripts/predictN2V.py'
       ],
 
+
       install_requires=[
           "numpy",
-          "scipy",
-          "matplotlib",
-          "six",
-          "keras>=2.1.1,<2.4.0",
-          "tifffile>=2020.5.11",
+          "tifffile",
           "imagecodecs>=2020.2.18",
-          "tqdm",
           "backports.tempfile;python_version<'3.4'",
-          "csbdeep>=0.6.0,<0.7.0",
+          "csbdeep>=0.7.2,<0.8.0",
           "Pillow",
           "ruamel.yaml>=0.16.10"
-        ]
+      ],
+
+      extras_require={
+          "testing": ["pytest"]
+      }
       )

tests/test_Noise2VoidDataGenerator.py

@@ -5,7 +5,6 @@
 
 
 def test_generate_patches_2D():
-
     if not os.path.isdir('data'):
         os.mkdir('data')
     zip_path = "data/RGB.zip"
@@ -25,8 +24,8 @@ def test_generate_patches_2D():
     patches = datagen.generate_patches_from_list(imgs, shape=(110, 280))
     assert len(patches) == 100
 
-def test_generate_patches_3D():
 
+def test_generate_patches_3D():
     if not os.path.isdir('data'):
         os.mkdir('data')
     zip_path = 'data/flywing-data.zip'
@@ -44,4 +43,3 @@ def test_generate_patches_3D():
     assert len(patches) == 1
     patches = datagen.generate_patches_from_list(imgs[:1], shape=(5, 52, 174))
     assert len(patches) == 210
-

tests/test_Noise2VoidDataWrapper.py

@@ -92,7 +92,7 @@ def create_data(y_shape):
         Y = np.random.rand(*y_shape)
         return Y
 
-    def random_neighbor_withCP_uniform(patch, coord, dims):
+    def random_neighbor_withCP_uniform(patch, coord, dims, structN2Vmask=None):
         return np.random.rand(1, 1)
 
     def _getitem2D(y_shape):