Merge branch 'master' into slides

Author: jonathanrocher

README.md

@@ -33,21 +33,38 @@ git clone git@github.com:jonathanrocher/ets_tutorial.git
 
 ### EDM users (recommended)
 First, download and install EDM from https://www.enthought.com/edm/. Then, 
-open a `Terminal`/`Powershell`/`Cmd Prompt`/ and navigate to the folder 
-where the repo was cloned. Enter the following command to create a 
-dedicated Python environment and install all dependencies in it:
+open a `Terminal`/`Powershell`/`Cmd Prompt`/ and create a lighweight bootstrap environment to run the installation commands.
 ```commandline
-python ci build
+edm envs create bootstrap
+edm install --environment bootstrap click
+```
+Next, enter the following commands to create a 
+dedicated Python environment called `ets_tutorial` and install all dependencies in it:
+```commandline
+edm run -e bootstrap -- python ci build --environment ets_tutorial
+```
+All application related `python` commands are assumed to run in this
+environment. You can activate the environment with:
+```commandline
+edm shell -e ets_tutorial
 ```
 
 ### Conda users
-[TODO]
+Create a new conda environment called `ets_tutorial` and install local
+dependencies with the following commands:
+```commandline
+conda create -n ets_tutorial python=3.6 pandas matplotlib traits traitsui scikit-image pillow pyqt ipython importlib_resources importlib_metadata
+conda activate ets_tutorial
+python setup.py install
+```
+Activate the new environment with `conda activate ets_tutorial`.
 
 ### pip users
 Assuming a Python environment is created and activated on your machine, for 
 example from https://www.python.org/, 
 ```commandline
 pip install pandas matplotlib traits traitsui scikits-image pillow pyqt5 ipython
+python setup.py install
 ```
 
 ## Getting help
@@ -78,11 +95,11 @@ are able to reach the end goal.
   - step 3: GUI: first traitsUI views
   - step 4: pyface application: tree navigator and double-click on an image to 
     display the traitsUI view of the image.
+  - INTERLUDE: code structure for scalability
   - step 5: Fuller pyface application: 
     - add folder editor to display a table of metadata for all images inside
     - add button to launch the face detection on all images
     - add widgets to filter images
-  - INTERLUDE: code structure for scalability
   - step 6: pyface application: adding menu and branding 
   - step 7: pyface application: advanced features [OPTIONAL]
   - step 8: 1-click installer

sample_images/20220121_080128.jpg

@@ -7,33 +7,50 @@
 from skimage.feature import Cascade
 import matplotlib.pyplot as plt
 from matplotlib import patches
-from os.path import join
+from os.path import join, splitext
 import numpy as np
 
 # ETS imports
-from traits.api import Dict, File, HasStrictTraits, List, observe
+from traits.api import (
+    Array, cached_property, Dict, File, HasStrictTraits, List, Property
+)
+
+SUPPORTED_FORMATS = [".png", ".jpg", ".jpeg", ".PNG", ".JPG", ".JPEG"]
 
 
 class ImageFile(HasStrictTraits):
     """ Model to hold an image file.
     """
     filepath = File
 
+    metadata = Property(Dict, depends_on="filepath")
+
+    data = Property(Array, depends_on="filepath")
+
     faces = List
 
-    metadata = Dict
+    def _is_valid_file(self):
+        return (
+            bool(self.filepath) and
+            splitext(self.filepath)[1].lower() in SUPPORTED_FORMATS
+        )
 
-    def to_array(self):
+    @cached_property
+    def _get_data(self):
+        if not self._is_valid_file():
+            return np.array([])
         with PIL.Image.open(self.filepath) as img:
             return np.asarray(img)
 
-    @observe("filepath")
-    def update_metadata(self, event):
-
+    @cached_property
+    def _get_metadata(self):
+        if not self._is_valid_file():
+            return {}
         with PIL.Image.open(self.filepath) as img:
             exif = img._getexif()
-        self.metadata = {TAGS[k]: v for k, v in exif.items()
-                         if k in TAGS}
+        if not exif:
+            return {}
+        return {TAGS[k]: v for k, v in exif.items() if k in TAGS}
 
     def detect_faces(self):
         # Load the trained file from the module root.
@@ -42,14 +59,13 @@ def detect_faces(self):
         # Initialize the detector cascade.
         detector = Cascade(trained_file)
 
-        detected = detector.detect_multi_scale(img=self.to_array(),
+        detected = detector.detect_multi_scale(img=self.data,
                                                scale_factor=1.2,
                                                step_ratio=1,
                                                min_size=(60, 60),
                                                max_size=(600, 600))
         self.faces = detected
-
-        self.metadata["Number of faces detected"] = len(detected)
+        return self.faces
 
 
 if __name__ == "__main__":
@@ -72,7 +88,7 @@ def detect_faces(self):
 
         # Visualize results ---------------------------------------------------
 
-        plt.imshow(img.to_array())
+        plt.imshow(img.data)
         img_desc = plt.gca()
         plt.set_cmap('gray')
 

sample_images/IMG-0325_xmas_2020.jpg

@@ -17,21 +17,19 @@ class TestImageFile(TestCase):
     def test_no_image_file(self):
         img = ImageFile()
         self.assertEqual(img.metadata, {})
-        data = img.to_array()
-        self.assertIsInstance(data, np.ndarray)
-        self.assertEqual(data.shape, (0,))
+        self.assertIsInstance(img.data, np.ndarray)
+        self.assertEqual(img.data.shape, (0,))
 
     def test_image_metadata(self):
         img = ImageFile(filepath=SAMPLE_IMG1)
         self.assertNotEqual(img.metadata, {})
         for key in ['ExifVersion', 'ExifImageWidth', 'ExifImageHeight']:
             self.assertIn(key, img.metadata.keys())
-        data = img.to_array()
         expected_shape = (img.metadata['ExifImageHeight'],
                           img.metadata['ExifImageWidth'], 3)
-        self.assertEqual(data.shape, expected_shape)
+        self.assertEqual(img.data.shape, expected_shape)
 
     def test_detects_faces(self):
         img = ImageFile(filepath=SAMPLE_IMG1)
+        img.detect_faces()
         self.assertEqual(len(img.faces), 5)
-        self.assertEqual(img.metadata["Number of faces"], 5)

stage0_starting_script/face_detect.py stage1_starting_script/face_detect.pystage0_starting_script/face_detect.py renamed to stage1_starting_script/face_detect.py

@@ -1,4 +1,5 @@
 # General imports
+from os.path import splitext
 import PIL.Image
 import matplotlib.pyplot as plt
 import numpy as np
@@ -9,23 +10,24 @@
 
 # ETS imports
 from traits.api import (
-    Dict,
-    File,
-    HasStrictTraits,
-    List,
-    observe,
+    Array, cached_property, Dict, File, HasStrictTraits, List,
+    Property
 )
 from traitsui.api import Item, OKButton, View
 
+SUPPORTED_FORMATS = [".png", ".jpg", ".jpeg", ".PNG", ".JPG", ".JPEG"]
+
 
 class ImageFile(HasStrictTraits):
     """ Model to hold an image file.
     """
     filepath = File
 
-    faces = List
+    metadata = Property(Dict, depends_on="filepath")
+
+    data = Property(Array, depends_on="filepath")
 
-    metadata = Dict
+    faces = List
 
     traits_view = View(
         Item(name='filepath', show_label=False),
@@ -34,58 +36,52 @@ class ImageFile(HasStrictTraits):
         width=640
     )
 
-    def to_array(self):
-        if not self.filepath:
-            return np.array([])
+    def _is_valid_file(self):
+        return (
+            bool(self.filepath) and
+            splitext(self.filepath)[1].lower() in SUPPORTED_FORMATS
+        )
 
+    @cached_property
+    def _get_data(self):
+        if not self._is_valid_file():
+            return np.array([])
         with PIL.Image.open(self.filepath) as img:
             return np.asarray(img)
 
-    @observe("filepath")
-    def _update_faces_and_metadata(self, event):
-        self.metadata = {}
-        self._update_metadata_with_exif()
-        self._detect_faces()
-        print(self.metadata)
-        print(f"Number of faces: {self.metadata['Number of faces']}")
-
-    def _update_metadata_with_exif(self):
-        if not self.filepath:
-            return
+    @cached_property
+    def _get_metadata(self):
+        if not self._is_valid_file():
+            return {}
         with PIL.Image.open(self.filepath) as img:
             exif = img._getexif()
         if not exif:
-            return
-        self.metadata.update(
-            {TAGS[k]: v for k, v in exif.items() if k in TAGS}
-        )
+            return {}
+        return {TAGS[k]: v for k, v in exif.items() if k in TAGS}
 
-    def _detect_faces(self):
-        self.faces = []
-        if not self.filepath:
-            return
+    def detect_faces(self):
         # Load the trained file from the module root.
         trained_file = data.lbp_frontal_face_cascade_filename()
+
         # Initialize the detector cascade.
         detector = Cascade(trained_file)
-        faces = detector.detect_multi_scale(
-            img=self.to_array(),
-            scale_factor=1.2,
-            step_ratio=1,
-            min_size=(60, 60),
-            max_size=(600, 600)
-        )
-        self.faces.extend(faces)
-        self.metadata['Number of faces'] = len(self.faces)
+
+        detected = detector.detect_multi_scale(img=self.data,
+                                               scale_factor=1.2,
+                                               step_ratio=1,
+                                               min_size=(60, 60),
+                                               max_size=(600, 600))
+        self.faces = detected
+        return self.faces
 
 
 if __name__ == '__main__':
     img = ImageFile()
     img.configure_traits()
 
-    plt.imshow(img.to_array())
+    plt.imshow(img.data)
     img_desc = plt.gca()
-    for patch in img.faces:
+    for patch in img.detect_faces():
         img_desc.add_patch(
             patches.Rectangle(
                 (patch['c'], patch['r']),

stage2.1_traited_script/traited_face_detect.py

@@ -19,24 +19,22 @@ class TestImageFile(TestCase):
     def test_no_image_file(self):
         img = ImageFile()
         self.assertEqual(img.metadata, {})
-        data = img.to_array()
-        self.assertIsInstance(data, np.ndarray)
-        self.assertEqual(data.shape, (0,))
+        self.assertIsInstance(img.data, np.ndarray)
+        self.assertEqual(img.data.shape, (0,))
 
     def test_image_metadata(self):
         img = ImageFile(filepath=SAMPLE_IMG1)
         self.assertNotEqual(img.metadata, {})
         for key in ['ExifVersion', 'ExifImageWidth', 'ExifImageHeight']:
             self.assertIn(key, img.metadata.keys())
-        data = img.to_array()
         expected_shape = (img.metadata['ExifImageHeight'],
                           img.metadata['ExifImageWidth'], 3)
-        self.assertEqual(data.shape, expected_shape)
+        self.assertEqual(img.data.shape, expected_shape)
 
     def test_detects_faces(self):
         img = ImageFile(filepath=SAMPLE_IMG1)
+        img.detect_faces()
         self.assertEqual(len(img.faces), 5)
-        self.assertEqual(img.metadata["Number of faces"], 5)
 
 
 class TestImageFileView(TestCase):