Allow annotation files with raster type (yet not supported) (#245)

Allow the reading of newer Darwin v7 annotations, Expand tests for annotations --------- Co-authored-by: Ajey Pai K <[email protected]>
NKI-AI · Aug 5, 2024 · 3a6a10f · 3a6a10f
1 parent 8996349
commit 3a6a10f
Show file tree

Hide file tree

Showing 5 changed files with 157 additions and 6 deletions.
diff --git a/dlup/annotations.py b/dlup/annotations.py
@@ -78,6 +78,7 @@ class AnnotationType(str, Enum):
     BOX = "BOX"
     POLYGON = "POLYGON"
     TAG = "TAG"
+    RASTER = "RASTER"
 
 
 class AnnotationSorting(str, Enum):
@@ -152,6 +153,9 @@ def _get_v7_metadata(filename: pathlib.Path) -> Optional[dict[str, DarwinV7Metad
     output = {}
     for sample in v7_metadata["classes"]:
         annotation_type = _v7_annotation_type_to_dlup_annotation_type(sample["type"])
+        # This is not implemented and can be skipped. The main function will raise a NonImplementedError
+        if annotation_type == AnnotationType.RASTER:
+            continue
 
         label = sample["name"]
         color = sample["color"][5:-1].split(",")
@@ -936,6 +940,9 @@ def from_darwin_json(
             annotation_type = _v7_annotation_type_to_dlup_annotation_type(
                 curr_annotation.annotation_class.annotation_type
             )
+            if annotation_type == AnnotationType.RASTER:
+                raise NotImplementedError("Raster annotations are not supported.")
+
             annotation_color = v7_metadata[name].color if v7_metadata else None
 
             if annotation_type == AnnotationType.TAG:
@@ -1241,4 +1248,7 @@ def _v7_annotation_type_to_dlup_annotation_type(annotation_type: str) -> Annotat
     if annotation_type == "tag":
         return AnnotationType.TAG
 
+    if annotation_type == "raster_layer":
+        return AnnotationType.RASTER
+
     raise NotImplementedError(f"annotation_type {annotation_type} is not implemented or not a valid dlup type.")
diff --git a/tests/files/.v7/metadata.json b/tests/files/.v7/metadata.json
@@ -0,0 +1,82 @@
+{
+  "classes": [
+    {
+      "name": "lymphocyte (cell)",
+      "type": "keypoint",
+      "description": null,
+      "color": "rgba(0,236,123,1.0)",
+      "sub_types": [
+        "inference"
+      ],
+      "properties": []
+    },
+    {
+      "name": "tumor (cell)",
+      "type": "bounding_box",
+      "description": null,
+      "color": "rgba(255,46,0,1.0)",
+      "sub_types": [
+        "inference"
+      ],
+      "properties": []
+    },
+    {
+      "name": "tumor (cell)",
+      "type": "keypoint",
+      "description": null,
+      "color": "rgba(255,92,0,1.0)",
+      "sub_types": [
+        "inference"
+      ],
+      "properties": []
+    },
+    {
+      "name": "lymphocyte (cell)",
+      "type": "bounding_box",
+      "description": null,
+      "color": "rgba(0,236,123,1.0)",
+      "sub_types": [
+        "inference"
+      ],
+      "properties": []
+    },
+    {
+      "name": "tumor (area)",
+      "type": "polygon",
+      "description": null,
+      "color": "rgba(255,46,0,1.0)",
+      "sub_types": [
+        "inference"
+      ],
+      "properties": []
+    },
+    {
+      "name": "stroma (area)",
+      "type": "polygon",
+      "description": null,
+      "color": "rgba(0,236,123,1.0)",
+      "sub_types": [
+        "inference"
+      ],
+      "properties": []
+    },
+    {
+      "name": "ROI (segmentation)",
+      "type": "bounding_box",
+      "description": null,
+      "color": "rgba(143,0,255,1.0)",
+      "sub_types": [
+        "inference"
+      ],
+      "properties": []
+    },
+    {
+      "name": "__raster_layer__",
+      "type": "raster_layer",
+      "description": "System class for all team's annotations with raster_layer type",
+      "color": null,
+      "sub_types": [],
+      "properties": []
+    }
+  ]
+}
diff --git a/tests/files/103S.json b/tests/files/103S.json
@@ -8,26 +8,26 @@
       "dataset": {
         "name": "PlayGround",
         "slug": "playground",
-        "dataset_management_url": "https://darwin.v7labs.com/datasets/669752/dataset-management"
+        "dataset_management_url": "dummy-url"
       },
       "item_id": "0189bfc8-4487-4b1f-5516-30150df25cc6",
       "team": {
-        "name": "Ellogon AI",
-        "slug": "ellogon-ai"
+        "name": "test-team",
+        "slug": "test-team"
       },
-      "workview_url": "https://darwin.v7labs.com/workview?dataset=669752&item=0189bfc8-4487-4b1f-5516-30150df25cc6"
+      "workview_url": "dummy-workview-url"
     },
     "slots": [
       {
         "type": "image",
         "slot_name": "0",
         "width": 63488,
         "height": 40064,
-        "thumbnail_url": "https://darwin.v7labs.com/api/v2/teams/ellogon-ai/files/1d10a9da-81c0-403f-98a4-98e65e5e80e4/thumbnail",
+        "thumbnail_url": "dummy-url",
         "source_files": [
           {
             "file_name": "103S.tif",
-            "url": "https://darwin.v7labs.com/api/v2/teams/ellogon-ai/uploads/65969e74-60b3-49b8-99de-c7af68a77860"
+            "url": "dummy-url"
           }
         ]
       }

diff --git a/tests/files/raster.json b/tests/files/raster.json
@@ -0,0 +1,52 @@
+{
+  "version": "2.0",
+  "schema_ref": "https://darwin-public.s3.eu-west-1.amazonaws.com/darwin_json/2.0/schema.json",
+  "item": {
+    "name": "dummy.tif",
+    "path": "/",
+    "source_info": {
+      "dataset": {
+        "name": "PlayGround",
+        "slug": "playground",
+        "dataset_management_url": "dummy-url"
+      },
+      "item_id": "0189bfc8-4487-4b1f-5516-30150df25cc6",
+      "team": {
+        "name": "test-team",
+        "slug": "test-team"
+      },
+      "workview_url": "dummy-url"
+    },
+    "slots": [
+      {
+        "type": "image",
+        "slot_name": "0",
+        "width": 63488,
+        "height": 40064,
+        "thumbnail_url": "dummy-url",
+        "source_files": [
+          {
+            "file_name": "103S.tif",
+            "url": "dummy-url"
+          }
+        ]
+      }
+    ]
+  },
+  "annotations": [
+    {
+      "raster_layer": {
+        "dense_rle":"placeholder",
+          "total_pixels": 1000,
+          "mask_annotation_ids_mapping": {
+            "0": "A random ID"
+          }
+      },
+      "id": "A random ID",
+      "name": "placeholder",
+      "slot_names": [
+        "0"
+      ]
+    }
+  ]
+}
diff --git a/tests/test_annotations.py b/tests/test_annotations.py
@@ -53,6 +53,7 @@ class TestAnnotations:
         geojson_annotations = WsiAnnotations.from_geojson([pathlib.Path(geojson_out.name)])
 
     _v7_annotations = None
+    _v7_raster_annotations = None
 
     @property
     def v7_annotations(self):
@@ -61,6 +62,12 @@ def v7_annotations(self):
             self._v7_annotations = WsiAnnotations.from_darwin_json(pathlib.Path(__file__).parent / "files/103S.json")
         return self._v7_annotations
 
+    def test_raster_annotations(self):
+        if self._v7_raster_annotations is None:
+            assert pathlib.Path(pathlib.Path(__file__).parent / "files/raster.json").exists()
+            with pytest.raises(NotImplementedError):
+                WsiAnnotations.from_darwin_json(pathlib.Path(__file__).parent / "files/raster.json")
+
     def test_conversion_geojson(self):
         # We need to read the asap annotations and compare them to the geojson annotations
         v7_region = self.v7_annotations.read_region((15300, 19000), 1.0, (2500.0, 2500.0))