Merge branch 'release_branch_1_5_0' into 'master'

Release branch 1 5 0

Closes #88

See merge request 3d/PandoraBox/pandora_plugins/plugin_libsgm!120

CHANGELOG.md

@@ -1,5 +1,14 @@
 # Changelog
 
+## 1.5.0a1 (Novembre 2023)
+
+### Changed
+- Adding check for geometric_prior option in check_conf function. [#88]
+- New format for disparity in the user configuration file. [#87]
+- Change read_img function to create_dataset_from_inputs. [#83]
+- Update with new API of Pandora run and compute_cost_volume function. [#94]
+
+
 ## 1.4.0 (April 2023)
 
 ### Added 

pandora_plugin_libsgm/abstract_sgm.py

@@ -54,13 +54,15 @@ class AbstractSGM(optimization.AbstractOptimization):
     _DIRECTIONS = [[0, 1], [1, 0], [1, 1], [1, -1], [0, -1], [-1, 0], [-1, -1], [-1, 1]]
     _USE_CONFIDENCE = False
 
-    def __init__(self, **cfg: Union[str, int, float, bool, dict]):
+    def __init__(self, img: xr.Dataset, **cfg: Union[str, int, float, bool, dict]):
         """
+        :param img: xarray.Dataset of image with metadata
+        :type img: xarray.Dataset
         :param cfg: optional configuration, {'P1': value, 'P2': value, 'alpha': value, 'beta': value, 'gamma': value,
                                             'p2_method': value}
         :type cfg: dict
         """
-        self.cfg = self.check_conf(**cfg)
+        self.cfg = self.check_conf(img, **cfg)
         self._sgm_version = self.cfg["sgm_version"]
         self._overcounting = self.cfg["overcounting"]
         self._min_cost_paths = self.cfg["min_cost_paths"]
@@ -78,10 +80,14 @@ def desc(self):
         Describes the optimization method
         """
 
-    def check_conf(self, **cfg: Union[str, int, float, bool, dict]) -> Dict[str, Union[str, int, float, bool, dict]]:
+    def check_conf(
+            self, img: xr.Dataset, **cfg: Union[str, int, float, bool, dict]
+        ) -> Dict[str, Union[str, int, float, bool, dict]]:
         """
         Add default values to the dictionary if there are missing elements and check if the dictionary is correct
 
+        :param img: xarray.Dataset of image with metadata
+        :type img: xarray.Dataset
         :param cfg: optimization configuration
         :type cfg: dict
         :return cfg: optimization configuration updated
@@ -101,6 +107,14 @@ def check_conf(self, **cfg: Union[str, int, float, bool, dict]) -> Dict[str, Uni
         if cfg["optimization_method"] == "sgm" and "geometric_prior" in cfg:
             logging.error("Geometric prior not available for SGM optimization")
             sys.exit(1)
+
+        if "geometric_prior" in cfg:
+            source = cfg["geometric_prior"]["source"]
+            if source in ["classif", "segm"] and not source in img.data_vars:
+                    logging.error(
+                        "For performing the 3SGM optimization step in the pipeline, left %s must be present.", source
+                    )
+                    sys.exit(1)
 
         schema = {
             "sgm_version": And(str, lambda x: is_method(x, ["c++", "python_libsgm", "python_libsgm_parall"])),

pandora_plugin_libsgm/lib_3sgm.py

@@ -47,13 +47,15 @@ class SEGSEMSGM(abstract_sgm.AbstractSGM):
     _GEOMETRIC_PRIOR = {"source": "internal"}
     _AVAILABLE_GEOMETRIC_PRIOR = ["internal", "classif", "segm"]
 
-    def __init__(self, **cfg: Union[str, int, float, bool]):
+    def __init__(self, img: xr.Dataset, **cfg: Union[str, int, float, bool]):
         """
+        :param img: xarray.Dataset of image with metadata
+        :type img: xarray.Dataset
         :param cfg: optional configuration, {}
         :type cfg: dict
         :return: None
         """
-        super().__init__(**cfg)
+        super().__init__(img, **cfg)
         self.cfg = self.check_geometric_prior(cfg)
         self._geometric_prior = self.cfg["geometric_prior"]
         if self._geometric_prior["source"] == "classif":  # type: ignore

setup.cfg

@@ -43,7 +43,7 @@ install_requires =
     numpy
     rasterio
     libsgm==0.4.*
-    pandora>=1.5.0
+    pandora==1.6.0a1
     xarray
     json-checker
 

tests/conftest.py

@@ -45,80 +45,86 @@ def configurations_path(resource_path_root):
 @pytest.fixture()
 def left_cones(inputs_path):
     """Cones images."""
-    path = inputs_path / "left.png"
-    return pandora.read_img(str(path), no_data=np.nan, mask=None)
+    return pandora.create_dataset_from_inputs(
+        {"img": str(inputs_path / "left.png"), "nodata": np.nan, "mask": None, "disp": [-60, 0]}
+    )
 
 
 @pytest.fixture()
 def right_cones(inputs_path):
     """Cones images."""
-    path = inputs_path / "right.png"
-    return pandora.read_img(str(path), no_data=np.nan, mask=None)
+    return pandora.create_dataset_from_inputs(
+        {"img": str(inputs_path / "right.png"), "nodata": np.nan, "mask": None, "disp": [0, 60]}
+    )
 
 
 @pytest.fixture()
 def left_rgb(inputs_path):
     """Cones images."""
-    path = inputs_path / "left_rgb.tif"
-    return pandora.read_img(str(path), no_data=np.nan, mask=None)
+    return pandora.create_dataset_from_inputs(
+        {"img": str(inputs_path / "left_rgb.tif"), "nodata": np.nan, "mask": None, "disp": [-60, 0]}
+    )
 
 
 @pytest.fixture()
 def right_rgb(inputs_path):
     """Cones images."""
-    path = inputs_path / "right_rgb.tif"
-    return pandora.read_img(str(path), no_data=np.nan, mask=None)
+    return pandora.create_dataset_from_inputs(
+        {"img": str(inputs_path / "right_rgb.tif"), "nodata": np.nan, "mask": None}
+    )
 
 
 @pytest.fixture()
 def left_cones_classif(inputs_path):
     """Cones images with classification."""
-    path = inputs_path / "left.png"
-    classif_path = inputs_path / "left_classif.tif"
-    return pandora.read_img(
-        str(path),
-        no_data=np.nan,
-        mask=None,
-        classif=str(classif_path),
+    return pandora.create_dataset_from_inputs(
+        {
+            "img": str(inputs_path / "left.png"),
+            "nodata": np.nan,
+            "mask": None,
+            "classif": str(inputs_path / "left_classif.tif"),
+            "disp": [-60, 0],
+        }
     )
 
 
 @pytest.fixture()
 def right_cones_classif(inputs_path):
     """Cones images with classification."""
-    path = inputs_path / "right.png"
-    classif_path = inputs_path / "right_classif.tif"
-    return pandora.read_img(
-        str(path),
-        no_data=np.nan,
-        mask=None,
-        classif=str(classif_path),
+    return pandora.create_dataset_from_inputs(
+        {
+            "img": str(inputs_path / "right.png"),
+            "nodata": np.nan,
+            "mask": None,
+            "classif": str(inputs_path / "right_classif.tif"),
+        }
     )
 
 
 @pytest.fixture()
 def left_cones_segm(inputs_path):
     """Cones images with segmentation."""
-    path = inputs_path / "left.png"
-    segment_path = inputs_path / "left_classif.tif"
-    return pandora.read_img(
-        str(path),
-        no_data=np.nan,
-        mask=None,
-        segm=str(segment_path),
+    return pandora.create_dataset_from_inputs(
+        {
+            "img": str(inputs_path / "left.png"),
+            "nodata": np.nan,
+            "mask": None,
+            "segm": str(inputs_path / "left_classif.tif"),
+            "disp": [-60, 0],
+        }
     )
 
 
 @pytest.fixture()
 def right_cones_segm(inputs_path):
     """Cones images with segmentation."""
-    path = inputs_path / "right.png"
-    segment_path = inputs_path / "right_classif.tif"
-    return pandora.read_img(
-        str(path),
-        no_data=np.nan,
-        mask=None,
-        segm=str(segment_path),
+    return pandora.create_dataset_from_inputs(
+        {
+            "img": str(inputs_path / "right.png"),
+            "nodata": np.nan,
+            "mask": None,
+            "segm": str(inputs_path / "right_classif.tif"),
+        }
     )