Implementation of Deep Learning unit tests, as well as paths to the 2d data for windows users in the UNet jupyter notebook.

Unit tests have been implemented for classes and functions that relate to Deep Learning tasks such as defining Models, Hyperparameters, Datasets / Dataloaders, as well as the training process.

The process is documented in the following notion file:
https://www.notion.so/qim-dtu/Unit-tests-to-DL-funtions-e8ad1fc1aa134da2ab26474cc6480ca2

the following scripts have been made under qim3d/tests/:

• models/test_unet.py (unit tests for UNet() and Hyperparameters())
• utils/test_augmentations.py (unit tests for Augmentations() and ValueErrors)
• utils/test_data.py (unit tests for Dataset() class, resizing of images with crop, padding, prepare_datasets() and prepare_dataloaders())
• utils/test_models.py (unit tests for output of model_summary() and inference())

Problem:

• Some deep learning tests require considerably more time to run (around 15s to 20s) especially when a Dataloader is involved. This is the case for prepare_dataloaders() as well as the model_summary(). Should they still be included?
• I couldn't think of a way to test the train_model() function, since that requires training a DL model, and would also take too much time.

docs/notebooks/annotation_tool.ipynb

+%% Cell type:markdown id: tags:
 
+# Image annotation tool
+This notebook shows how the annotation interface can be used to create masks for images
 
+%% Cell type:code id: tags:
 
+``` python
+import qim3d
+import matplotlib.pyplot as plt
+import matplotlib as mpl
+import numpy as np
+%matplotlib inline
 ```
 
+%% Cell type:code id: tags:
 
+``` python
+# Load 2D example image
+img = qim3d.examples.blobs_256x256
+
+# Display image
+plt.imshow(img)
+plt.show()
 ```
 
+%% Cell type:code id: tags:
 
+``` python
+# Start annotation tool
+interface = qim3d.gui.annotation_tool.Interface()
+interface.max_masks = 4
+
+# We can directly pass the image we loaded to the interface
+interface.launch(img=img)
 ```
 
+%% Cell type:code id: tags:
 
+``` python
+# When 'prepare mask for download' is pressed once, the mask can be retrieved with the get_result() method
+mask = interface.get_result()
 ```
 
+%% Cell type:markdown id: tags:
 
+## Check the obtained mask
 
+%% Cell type:code id: tags:
 
+``` python
+print (f"Original image shape..: {img.shape}")
+print (f"Mask image shape......: {mask.shape}")
+print (f"\nNumber of masks: {np.max(mask)}")
 ```
 
+%% Cell type:markdown id: tags:
 
+## Show the masked regions
 
+%% Cell type:code id: tags:
 
+``` python
+%matplotlib inline
+
+nmasks = np.max(mask)
+fig, axs = plt.subplots(nrows=1, ncols=nmasks+2, figsize=(12,3))
+
+# Show original image
+axs[0].imshow(img)
+axs[0].set_title("Original")
+axs[0].axis('off')
+
+
+# Show masks
+cmap = mpl.colormaps["rainbow"].copy()
+cmap.set_under(color='black') # Sets the background to black
+axs[1].imshow(mask, interpolation='none', cmap=cmap, vmin=1, vmax=nmasks+1)
+axs[1].set_title("Masks")
+axs[1].axis('off')
+
+# Show masked regions
+for idx in np.arange(2, nmasks+2):
+    mask_id = idx-1
+    submask = mask.copy()
+    submask[submask != mask_id] = 0
+
+    masked_img = img.copy()
+    masked_img[submask==0] = 0
+    axs[idx].imshow(masked_img)
+    axs[idx].set_title(f"Mask {mask_id}")
+
+    axs[idx].axis('off')
+
+plt.show()
 ```