Added GT to pipeline

.github/workflows/pipeline_run.yml

@@ -21,4 +21,4 @@ jobs:
         source env/bin/activate
         pip install -e .
         echo "Running pipeline"
-        boreholes-extract-all -l -i example/example_borehole_profile.pdf -o example/ -p example/predictions.json -m example/metadata.json
+        boreholes-extract-all -l -i example/example_borehole_profile.pdf -o example/ -p example/predictions.json -m example/metadata.json -g example/example_ground_truth.json

example/example_groundtruth.json

@@ -0,0 +1,18 @@
+{
+    "example_borehole_profile.pdf": {
+        "groundwater": [],
+        "layers": [],
+        "metadata": {
+            "coordinates": {
+                "E": 615790,
+                "N": 157500
+            },
+            "drilling_date": "1995-09-03",
+            "drilling_methods": null,
+            "original_name": "",
+            "project_name": "",
+            "reference_elevation": 788.6,
+            "total_depth": null
+        }
+    }
+}

src/stratigraphy/benchmark/ground_truth.py

@@ -16,8 +16,11 @@ class GroundTruth:
     def __init__(self, path: Path):
         self.ground_truth = defaultdict(dict)
 
-        with open(path) as in_file:
+        # Load the ground truth data
+        with open(path, encoding="utf-8") as in_file:
             ground_truth = json.load(in_file)
+
+        # Parse the ground truth data
         for borehole_profile, ground_truth_item in ground_truth.items():
             layers = ground_truth_item["layers"]
             self.ground_truth[borehole_profile]["layers"] = [
@@ -42,6 +45,6 @@ def for_file(self, file_name: str) -> dict:
         """
         if file_name in self.ground_truth:
             return self.ground_truth[file_name]
-        else:
-            logger.warning(f"No ground truth data found for {file_name}.")
-            return {}
+
+        logger.warning("No ground truth data found for %s.", file_name)
+        return {}