Most Configuration via Hydra

.github/workflows/tests.yml

@@ -15,7 +15,7 @@ jobs:
             fail-fast: false
             matrix:
                 os: [ubuntu-latest, windows-latest, macos-latest]
-                python-version: ['3.9', '3.10', '3.11', '3.12']
+                python-version: ['3.10', '3.11', '3.12']
                 exclude:
                     - os: windows-latest
                       python-version: '3.12'

.gitignore

@@ -152,8 +152,7 @@ lightning_logs/
 **/results/*
 **/cache/*
 **/outputs/*
-tests/tmp/cache/*
-tests/tmp/results/*
+tests/tmp/*
 user_cfgs/*
 
 inbuilt_cfgs/config.yml
@@ -162,8 +161,14 @@ inbuilt_cfgs/mf_config.yml
 .vscode/settings.json
 
 # Exclusions
+!docs/
+docs/*
+!docs/images/
+docs/images/*
 !docs/images/*.png
+
 !tests/fixtures/data/**/*.png
 !tests/fixtures/data/**/*.txt
 !tests/fixtures/data/**/*.csv
 !tests/fixtures/data/**/*.tif
+!tests/fixtures/data/**/*.json

.pre-commit-config.yaml

@@ -18,7 +18,7 @@ repos:
           - id: requirements-txt-fixer
 
     - repo: https://github.com/psf/black
-      rev: 24.4.2
+      rev: 24.8.0
       hooks:
           - id: black
 

README.md

@@ -57,7 +57,7 @@ datasets with upcoming support for [torchvision](https://pytorch.org/vision/stab
 
 Required Python modules for `minerva` are stated in the `setup.cfg`.
 
-`minerva` currently only supports `python` 3.9 -- 3.12.
+`minerva` currently only supports `python` 3.10 -- 3.12.
 
 <p align="right">(<a href="#top">back to top</a>)</p>
 

minerva/datasets/collators.py

@@ -39,45 +39,36 @@
 # =====================================================================================================================
 #                                                     IMPORTS
 # =====================================================================================================================
-from typing import Any, Callable, Dict, Iterable, Optional, Tuple
+from typing import Any, Callable, Iterable
 
+from hydra.utils import get_method
 from torchgeo.datasets.utils import stack_samples
 
-from minerva.utils import utils
-
 
 # =====================================================================================================================
 #                                                     METHODS
 # =====================================================================================================================
 def get_collator(
-    collator_params: Optional[Dict[str, str]] = None
+    collator_target: str = "torchgeo.datasets.stack_samples",
 ) -> Callable[..., Any]:
     """Gets the function defined in parameters to collate samples together to form a batch.
 
     Args:
-        collator_params (dict[str, str]): Optional; Dictionary that must contain keys for
-            ``'module'`` and ``'name'`` of the collation function. Defaults to ``config['collator']``.
+        collator_target (str): Dot based import path for collator method.
+            Defaults to :meth:`torchgeo.datasets.stack_samples`
 
     Returns:
-        ~typing.Callable[..., ~typing.Any]: Collation function found from parameters given.
+        ~typing.Callable[..., ~typing.Any]: Collation function found from target path given.
     """
     collator: Callable[..., Any]
-    if collator_params is not None:
-        module = collator_params.pop("module", "")
-        if module == "":
-            collator = globals()[collator_params["name"]]
-        else:
-            collator = utils.func_by_str(module, collator_params["name"])
-    else:
-        collator = stack_samples
-
+    collator = get_method(collator_target)
     assert callable(collator)
     return collator
 
 
 def stack_sample_pairs(
-    samples: Iterable[Tuple[Dict[Any, Any], Dict[Any, Any]]]
-) -> Tuple[Dict[Any, Any], Dict[Any, Any]]:
+    samples: Iterable[tuple[dict[Any, Any], dict[Any, Any]]]
+) -> tuple[dict[Any, Any], dict[Any, Any]]:
     """Takes a list of paired sample dicts and stacks them into a tuple of batches of sample dicts.
 
     Args:

minerva/datasets/dfc.py

@@ -38,7 +38,7 @@
 # =====================================================================================================================
 from glob import glob
 from pathlib import Path
-from typing import Any, Callable, Dict, List, Optional, Tuple
+from typing import Any, Callable, Optional
 
 import numpy as np
 import pandas as pd
@@ -84,7 +84,7 @@
     S2_BANDS_MR = [5, 6, 7, 9, 12, 13]
     S2_BANDS_LR = [1, 10, 11]
 
-    splits: List[str] = []
+    splits: list[str] = []
 
     igbp = False
 
@@ -133,13 +133,13 @@
         # Make sure parent dir exists.
         assert self.root.exists()
 
-        self.samples: List[Dict[str, str]]
+        self.samples: list[dict[str, str]]
 
     def load_sample(
         self,
-        sample: Dict[str, str],
+        sample: dict[str, str],
         index: int,
-    ) -> Dict[str, Any]:
+    ) -> dict[str, Any]:
         """Util function for reading data from single sample.
 
         Args:
@@ -192,7 +192,7 @@
             n_inputs += 2
         return n_inputs
 
-    def get_display_channels(self) -> Tuple[List[int], int]:
+    def get_display_channels(self) -> tuple[list[int], int]:
         """Select channels for preview images.
 
         Returns:
@@ -303,7 +303,7 @@
 
         return lc
 
-    def __getitem__(self, index: int) -> Dict[str, Any]:
+    def __getitem__(self, index: int) -> dict[str, Any]:
         """Get a single example from the dataset"""
 
         # Get and load sample from index file.
@@ -393,11 +393,11 @@
 
     def plot(
         self,
-        sample: Dict[str, Tensor],
+        sample: dict[str, Tensor],
         show_titles: bool = True,
         suptitle: Optional[str] = None,
-        classes: Optional[Dict[int, str]] = None,
-        colours: Optional[Dict[int, str]] = None,
+        classes: Optional[dict[int, str]] = None,
+        colours: Optional[dict[int, str]] = None,
     ) -> Figure:
         """Plot a sample from the dataset.
 
@@ -420,10 +420,15 @@
         # Reorder image from BGR to RGB.
         from kornia.color import BgrToRgb
 
+        from minerva.transforms import AdjustGamma
+
         bgr_to_rgb = BgrToRgb()
+        adjust_gamma = AdjustGamma(gamma=0.8)
+
+        # Reorder channels from BGR to RGB and adjust the gamma ready for plotting.
+        image = adjust_gamma(bgr_to_rgb(sample["image"][:3])).permute(1, 2, 0).numpy()
 
-        image = bgr_to_rgb(sample["image"][:3])
-        image = image.permute(1, 2, 0).numpy()
+        block_size_factor = 8
 
         # Use inbuilt class colours and classes mappings.
         if classes is None or colours is None:
@@ -454,30 +459,72 @@
 
         # Plot the image.
         axs[0].imshow(image)
-        axs[0].axis("off")
+
+        # Sets tick intervals to block size.
+        axs[0].set_xticks(
+            np.arange(0, image.shape[0] + 1, image.shape[0] // block_size_factor)
+        )
+        axs[0].set_yticks(
+            np.arange(0, image.shape[1] + 1, image.shape[1] // block_size_factor)
+        )
+
+        # Add grid overlay.
+        axs[0].grid(which="both", color="#CCCCCC", linestyle=":")
 
         # Plot the ground truth mask and predicted mask.
         mask_plot: Axes
         if showing_mask:
             mask_plot = axs[1].imshow(
                 mask, cmap=cmap, vmin=vmin, vmax=vmax, interpolation="none"
             )
-            axs[1].axis("off")
+
+            # Sets tick intervals to block size.
+            axs[1].set_xticks(
+                np.arange(0, mask.shape[0] + 1, mask.shape[0] // block_size_factor)
+            )
+            axs[1].set_yticks(
+                np.arange(0, mask.shape[1] + 1, mask.shape[1] // block_size_factor)
+            )
+
+            # Add grid overlay.
+            axs[1].grid(which="both", color="#CCCCCC", linestyle=":")
+
             if showing_prediction:
                 axs[2].imshow(
                     pred, cmap=cmap, vmin=vmin, vmax=vmax, interpolation="none"
                 )
-                axs[2].axis("off")
+
+                # Sets tick intervals to block size.
+                axs[2].set_xticks(
+                    np.arange(0, pred.shape[0] + 1, pred.shape[0] // block_size_factor)
+                )
+                axs[2].set_yticks(
+                    np.arange(0, pred.shape[1] + 1, pred.shape[1] // block_size_factor)
+                )
+
+                # Add grid overlay.
+                axs[2].grid(which="both", color="#CCCCCC", linestyle=":")
+
         elif showing_prediction:
             mask_plot = axs[1].imshow(
                 pred, cmap=cmap, vmin=vmin, vmax=vmax, interpolation="none"
             )
-            axs[1].axis("off")
+
+            # Sets tick intervals to block size.
+            axs[1].set_xticks(
+                np.arange(0, pred.shape[0] + 1, pred.shape[0] // block_size_factor)
+            )
+            axs[1].set_yticks(
+                np.arange(0, pred.shape[1] + 1, pred.shape[1] // block_size_factor)
+            )
+
+            # Add grid overlay.
+            axs[1].grid(which="both", color="#CCCCCC", linestyle=":")
 
         if showing_mask or showing_prediction:
             # Plots colour bar onto figure.
             clb = fig.colorbar(
-                mask_plot,
+                mask_plot,  # type: ignore[arg-type]
                 ax=axs,
                 location="top",
                 ticks=np.arange(0, len(colours)),