Merge branch 'css-materialdesign' of https://github.com/AaltoPML/VAI-Lab

src/vai_lab/DataProcessing/plugins/RGBcalculation.py

@@ -0,0 +1,66 @@
+import numpy as np
+from sklearn.base import BaseEstimator
+from vai_lab._plugin_templates import DataProcessingT
+
+_PLUGIN_READABLE_NAMES = {"RGBcalculation":"default"}  # type:ignore
+_PLUGIN_MODULE_OPTIONS = {"Type": "Other"}            # type:ignore
+_PLUGIN_REQUIRED_SETTINGS = {"Data": "str"}            # type:ignore
+_PLUGIN_OPTIONAL_SETTINGS = {}                         # type:ignore
+_PLUGIN_REQUIRED_DATA = {}                             # type:ignore
+_PLUGIN_OPTIONAL_DATA = {"X","Y","X_tst", 'Y_tst'}     # type:ignore
+
+class RGBcalculation(DataProcessingT):
+
+    def __init__(self):
+        """Initialises parent class. 
+            Passes `globals` dict of all current variables
+        """
+        super().__init__(globals())
+        self.proc = model()
+
+    # def configure(self, config: dict):
+    #     """Sets and parses plugin configurations options
+    #     :param config: dict of internal tags set in the XML config file 
+    #     """
+    #     super().configure(config)
+
+    # def set_data_in(self, data_in):
+    #     """Sets and parses incoming data
+    #     :param data_in: saves data as class variable
+    #                     expected type: aidesign.Data.Data_core.Data
+    #     """
+    #     super().set_data_in(data_in)
+
+    # def fit(self):
+    #     cleaned_options = self._clean_solver_options()
+    #     self.proc.set_params(**cleaned_options)
+    #     self.proc.fit(self.X)
+
+    # def transform(self, data: DataInterface) -> DataInterface:
+    #     data.append_data_column("X", pd.DataFrame(self.proc.transform(data)))
+    #     # if self.X_tst is not None:
+    #     #     data.append_data_column("X_test", pd.DataFrame(self.proc.transform(self.X_tst)))
+    #     print(data)
+    #     return data
+
+class model(BaseEstimator):
+    def __init__(self):
+        return
+
+    def fit(self, X):
+        return self
+
+    def transform(self, X):
+        """ Calculates the mean RGB value of an image.
+        
+        Parameters
+        ----------
+        X : {array-like, sparse matrix} of shape (pixels, pixels, RGB)
+            The image data with the  RGB information.
+        Returns
+        -------
+        mean_RGB : {int, float}
+            Mean RGB value of the input image.
+        """
+        mean_RGB = np.mean(X)
+        return mean_RGB

src/vai_lab/DataProcessing/plugins/RectangleDetection.py

@@ -0,0 +1,124 @@
+import numpy as np
+from sklearn.base import BaseEstimator
+from vai_lab._plugin_templates import DataProcessingT
+import matplotlib.pyplot as plt
+
+_PLUGIN_READABLE_NAMES = {"RectangleDetection":"default",
+                          "RectDet": "alias",            
+                          "rectangledetection": "alias"}    # type:ignore
+_PLUGIN_MODULE_OPTIONS = {"Type": "Other"}                  # type:ignore
+_PLUGIN_REQUIRED_SETTINGS = {"Data": "str"}                 # type:ignore
+_PLUGIN_OPTIONAL_SETTINGS = {"r": "int", 
+                             "c": "int", 
+                             "h":"float", 
+                             "w": "float"}                  # type:ignore
+_PLUGIN_REQUIRED_DATA = {}                                  # type:ignore
+_PLUGIN_OPTIONAL_DATA = {"X","Y","X_tst","Y_tst"}           # type:ignore
+
+
+class RectangleDetection(DataProcessingT):
+    """
+    Red rectangle detection in an image
+    """
+    def __init__(self):
+        """Initialises parent class. 
+            Passes `globals` dict of all current variables
+        """
+        super().__init__(globals())
+        self.proc = model()
+
+    # def configure(self, config: dict):
+    #     """Sets and parses plugin configurations options
+    #     :param config: dict of internal tags set in the XML config file 
+    #     """
+    #     super().configure(config)
+
+    # def set_data_in(self, data_in):
+    #     """Sets and parses incoming data
+    #     :param data_in: saves data as class variable
+    #                     expected type: aidesign.Data.Data_core.Data
+    #     """
+    #     super().set_data_in(data_in)
+
+    # def fit(self):
+    #     cleaned_options = self._clean_solver_options()
+    #     self.proc.set_params(**cleaned_options)
+    #     """
+    #     # TODO: Temporary solution until input data module is cofigured."""
+
+    #     # dst_img = os.path.join(
+    #     #     get_lib_parent_dir(),
+    #     #     'examples',
+    #     #     'crystalDesign')
+    #     # #iterating over dst_image to get the images as arrays
+    #     # # for image in sorted(os.listdir(dst_img)):
+    #     # arr = np.array(Image.open(os.path.join(dst_img, '00007.png')))
+    #     # self.proc.fit(arr)
+    #     """"""
+    #     self.proc.fit(self.X)
+
+    # def transform(self, data: DataInterface) -> DataInterface:
+    #     """
+    #     # TODO: Temporary solution until input data module is cofigured."""
+    #     # dst_img = os.path.join(
+    #     #     get_lib_parent_dir(),
+    #     #     'examples',
+    #     #     'crystalDesign')
+    #     # arr = np.array(Image.open(os.path.join(dst_img, '00007.png')))
+    #     # self.proc.transform(arr)
+    #     """"""
+    #     data.append_data_column("X", pd.DataFrame(self.proc.transform(data)))
+    #     return data
+
+class model(BaseEstimator):
+    def __init__(self, optional=False):
+        self.optional = optional
+
+    def fit(self, X, r=7, c=4, h=35, w=79, hs=34, vs=23, buffer=5):
+        """Defines the number of samples in the design, their placement
+        and their size.
+        
+        Parameters
+        ----------
+        X : {array-like, sparse matrix} of shape (pixels, pixels, RGB)
+            The image data with the samples with RGB information.
+            Samples are expected to be marked with a red rectangle.
+        r : int
+            Number of rows with samples.
+        c : int
+            Number of columns with samples.
+        h : float
+            Sample height.
+        w : int
+            Sample width.
+        hs : float
+            Horizontal spacing.
+        vs : int
+            Vertical spacing.
+        buffer : int
+            Pixel buffer. Inidcates the margin for the inner rectangle.
+        Returns
+        -------
+        self : object
+            Fitted model.
+        """
+        self.r = r
+        self.c = c
+        self.h = h
+        self.w = w
+        self.hs = hs
+        self.vs = vs
+        self.buffer = buffer
+        mask = (X[:,:,0] > 120) * (X[:,:,1] < 80) * (X[:,:,2] < 80)
+        self.ii_ini, self.jj_ini = np.unravel_index(mask.argmax(), mask.shape)
+        # return self
+
+    def transform(self, X):
+        X_dict = {}
+        fig, axs = plt.subplots(self.r, self.c)
+        for i in np.arange(self.r):
+            for j in np.arange(self.c):
+                X_dict[i,j] = X[self.ii_ini+self.h*i+self.vs*i+self.buffer*(1+i):self.ii_ini+self.h*(i+1)+self.vs*i-self.buffer*(1-i), 
+                                self.jj_ini+self.w*j+self.hs*j+self.buffer*(1+j):self.jj_ini+self.w*(j+1)+self.hs*j-self.buffer*(1-j),:]
+                axs[i, j].imshow(X_dict[i,j])
+        return X_dict

src/vai_lab/DataProcessing/plugins/argopt.py

@@ -0,0 +1,35 @@
+from distutils.command.config import config
+from numpy import argmin, argmax
+from vai_lab._plugin_templates import DataProcessingT
+import pandas as pd
+
+_PLUGIN_READABLE_NAMES = {"argopt": "default",
+                          "argmax": "alias",            
+                          "argmin": "alias"}            # type:ignore
+_PLUGIN_MODULE_OPTIONS = {"Type": "math operator"}      # type:ignore
+_PLUGIN_REQUIRED_SETTINGS = {"Data": "str"}             # type:ignore
+_PLUGIN_OPTIONAL_SETTINGS = {'min/max': "str"}          # type:ignore
+_PLUGIN_REQUIRED_DATA = {}                              # type:ignore
+_PLUGIN_OPTIONAL_DATA = {"X","Y","X_tst", 'Y_tst'}      # type:ignore
+
+class argopt(DataProcessingT):
+    """
+    Calculate the optimum argument
+    """
+
+    def __init__(self):
+        """Initialises parent class. 
+            Passes `globals` dict of all current variables
+        """
+        super().__init__(globals())
+
+    def fit(self):
+        self.cleaned_options = self._clean_solver_options()
+        return
+
+    def transform(self,data):
+        if config['min/max'] == 'max':
+            data.append_data_column("X", pd.DataFrame(argmax(self.X)))
+        else:
+            data.append_data_column("X", pd.DataFrame(argmin(self.X)))
+        return data

src/vai_lab/DataProcessing/plugins/imgtofeatvect.py

@@ -0,0 +1,51 @@
+from vai_lab._plugin_templates import DataProcessingT
+from sklearn.base import BaseEstimator
+import pandas as pd
+import numpy as np
+
+_PLUGIN_READABLE_NAMES = {"ImgToFeatVect":"default",
+                          "imgtofeatvect": "alias", 
+                          "ImageToFeatureVector": "alias"}  # type:ignore
+_PLUGIN_MODULE_OPTIONS = {"Type": "Other"}                  # type:ignore
+_PLUGIN_REQUIRED_SETTINGS = {"Data": "str"}                 # type:ignore
+_PLUGIN_OPTIONAL_SETTINGS = {}                              # type:ignore
+_PLUGIN_REQUIRED_DATA = {}                                  # type:ignore
+_PLUGIN_OPTIONAL_DATA = {"X","Y","X_tst","Y_tst"}           # type:ignore
+
+
+class ImgToFeatVect(DataProcessingT):
+
+    def __init__(self):
+        """Initialises parent class. 
+            Passes `globals` dict of all current variables
+        """
+        super().__init__(globals())
+        self.proc = model()
+
+
+class model(BaseEstimator):
+    def __init__(self, optional=False):
+        self.optional = optional
+
+    def fit(self, X):
+        """Defines the number of samples in the design, their placement
+        and their size.
+        
+        Parameters
+        ----------
+        X : {dict}
+            A dictionary containing images in each entry.
+        Returns
+        -------
+        self : object
+            Fitted model.
+        """
+        self.n = len(X.keys())
+        h,w,self.rgb = next(iter(X.values())).shape
+        self.d = h*w
+        return self
+
+    def transform(self, X):
+        X_mat = np.dstack([X[el].reshape(self.d,self.rgb) for el in X.keys()])
+        X_mat = X_mat.reshape(self.n,self.d,self.rgb)
+        return pd.DataFrame(X_mat)

src/vai_lab/DataProcessing/plugins/integral.py

@@ -0,0 +1,32 @@
+from scipy.integrate import simps as model
+from vai_lab._plugin_templates import DataProcessingT
+import pandas as pd
+
+_PLUGIN_READABLE_NAMES = {"Integral": "default",
+                          "integral": "alias"}          # type:ignore
+_PLUGIN_MODULE_OPTIONS = {"Type": "math operator"}      # type:ignore
+_PLUGIN_REQUIRED_SETTINGS = {"Data": "str"}             # type:ignore
+_PLUGIN_OPTIONAL_SETTINGS = {"dx": "float",
+                             "axis": "int",
+                             "even": "str"}             # type:ignore
+_PLUGIN_REQUIRED_DATA = {}                              # type:ignore
+_PLUGIN_OPTIONAL_DATA = {"X","Y","X_tst","Y_tst"}       # type:ignore
+
+class Integral(DataProcessingT):
+    """
+    Calculate integral of array using the composite trapezoidal rule
+    """
+
+    def __init__(self):
+        """Initialises parent class. 
+            Passes `globals` dict of all current variables
+        """
+        super().__init__(globals())
+        self.proc = model()
+
+    def fit(self):
+        return
+
+    def transform(self,data):
+        data.append_data_column("X", pd.DataFrame(self.proc(self.X)))
+        return data

src/vai_lab/UserInteraction/plugins/OptimisationInput.py

@@ -20,7 +20,7 @@
 _PLUGIN_MODULE_OPTIONS = {"layer_priority": 2,
                           "required_children": None}            # type:ignore
 _PLUGIN_REQUIRED_SETTINGS = {}                                  # type:ignore
-_PLUGIN_OPTIONAL_SETTINGS = {"Bounds": "list"}                      # type:ignore
+_PLUGIN_OPTIONAL_SETTINGS = {"Bounds": "list"}                  # type:ignore
 _PLUGIN_REQUIRED_DATA = {"X"}                                   # type:ignore
 
 

src/vai_lab/examples/xml_files/BO_demo.xml

@@ -0,0 +1,32 @@
+<pipeline>
+    <Initialiser name="Initialiser">
+        <inputdata>            
+            <X file="./examples/optimisation/X.csv" />
+        </inputdata>
+        <relationships>
+            <child name="User Interaction" />
+        </relationships>
+        <coordinates>
+            [(350.0,50),0,{}]
+        </coordinates>
+    </Initialiser>
+    <UserInteraction name="User Interaction">
+        <relationships>
+            <parent name="Initialiser" />
+            <child name="Output" />
+        </relationships>
+        <coordinates>
+            [(350.0,350.0),2,{0:'d0-u2'}]
+        </coordinates>
+        <plugin type="optimisationUI" />
+    </UserInteraction>
+    <Output name="Output">
+        <plugin type="" />
+        <relationships>
+            <parent name="User Interaction" />
+        </relationships>
+        <coordinates>
+            [(350.0,650),1,{2:'d2-u1'}]
+        </coordinates>
+    </Output>
+</pipeline>

src/vai_lab/examples/xml_files/DecisionMaking_demo.xml

@@ -0,0 +1,29 @@
+<pipeline>
+    <Initialiser name="Initialiser">
+        <relationships>
+            <child name="Decision Making" />
+        </relationships>
+        <coordinates>
+            [(350.0,50),0,{}]
+        </coordinates>
+    </Initialiser>
+    <DecisionMaking name="Decision Making">
+        <plugin type="" />
+        <relationships>
+            <parent name="Initialiser" />
+            <child name="Output" />
+        </relationships>
+        <coordinates>
+            [(350.0, 350.0), 2, {0: 'd0-u2'}]
+        </coordinates>
+    </DecisionMaking>
+    <Output name="Output">
+        <plugin type="" />
+        <relationships>
+            <parent name="Decision Making" />
+        </relationships>
+        <coordinates>
+            [(350.0, 650), 1, {2: 'd2-u1'}]
+        </coordinates>
+    </Output>
+</pipeline>