diff --git a/glmocr/config.py b/glmocr/config.py
index 789e206..00211a9 100644
--- a/glmocr/config.py
+++ b/glmocr/config.py
@@ -198,6 +198,7 @@ class LayoutConfig(_BaseConfig):
     layout_merge_bboxes_mode: Union[str, Dict[int, str]] = "large"
     label_task_mapping: Optional[Dict[str, Any]] = None
     use_polygon: bool = False
+    id2label: Optional[Dict[Union[int, str], str]] = None
 
     @field_validator("device")
     @classmethod
diff --git a/glmocr/layout/layout_detector.py b/glmocr/layout/layout_detector.py
index c4f1680..f5e96ed 100644
--- a/glmocr/layout/layout_detector.py
+++ b/glmocr/layout/layout_detector.py
@@ -77,8 +77,12 @@ def start(self):
             self._device = "cpu"
         self._model = self._model.to(self._device)
         if self.id2label is None:
-            self.id2label = self._model.config.id2label
-
+            self.id2label = getattr(self._model.config, "id2label", None)
+        if self.id2label is None:
+            raise RuntimeError(
+                "Missing id2label in both layout config and model config; "
+                "please set pipeline.layout.id2label."
+            )
         # Patch upstream _extract_polygon_points_by_masks to guard against
         # empty mask crops that crash cv2.resize with !ssize.empty().
         def _safe_extract(boxes, masks, scale_ratio):
@@ -126,6 +130,11 @@ def _safe_extract(boxes, masks, scale_ratio):
             return polygon_points
 
         self._image_processor._extract_polygon_points_by_masks = _safe_extract
+        if self.label_task_mapping is None:
+            logger.warning(
+                "layout.label_task_mapping is missing; defaulting all labels to text"
+            )
+            self.label_task_mapping = {"text": list(self.id2label.values())}
         logger.debug(f"PP-DocLayoutV3 loaded on device: {self._device}")
 
     def stop(self):
diff --git a/glmocr/tests/test_unit.py b/glmocr/tests/test_unit.py
index 62ba50d..6a01f93 100644
--- a/glmocr/tests/test_unit.py
+++ b/glmocr/tests/test_unit.py
@@ -208,6 +208,66 @@ def test_detector_device_selection_auto_cuda(self):
 
         assert det._device == "cuda:1"
 
+    def test_detector_defaults_label_task_mapping_from_model_id2label(self):
+        """Missing label_task_mapping falls back to a text bucket from id2label."""
+        self._require_layout_runtime()
+        from glmocr.config import LayoutConfig
+        from glmocr.layout.layout_detector import PPDocLayoutDetector
+
+        cfg = LayoutConfig(model_dir="dummy", device="cpu")
+        det = PPDocLayoutDetector(cfg)
+
+        mock_model = MagicMock()
+        mock_model.to = MagicMock(return_value=mock_model)
+        mock_model.eval = MagicMock()
+        mock_model.config = MagicMock()
+        mock_model.config.id2label = {0: "text", 1: "title"}
+        mock_proc = MagicMock()
+
+        with (
+            patch(
+                "glmocr.layout.layout_detector.PPDocLayoutV3ForObjectDetection.from_pretrained",
+                return_value=mock_model,
+            ),
+            patch(
+                "glmocr.layout.layout_detector.PPDocLayoutV3ImageProcessorFast.from_pretrained",
+                return_value=mock_proc,
+            ),
+        ):
+            det.start()
+
+        assert det.id2label == {0: "text", 1: "title"}
+        assert det.label_task_mapping == {"text": ["text", "title"]}
+
+    def test_detector_raises_when_id2label_missing_everywhere(self):
+        """Missing id2label in both config and model config raises a clear error."""
+        self._require_layout_runtime()
+        from glmocr.config import LayoutConfig
+        from glmocr.layout.layout_detector import PPDocLayoutDetector
+
+        cfg = LayoutConfig(model_dir="dummy", device="cpu")
+        det = PPDocLayoutDetector(cfg)
+
+        mock_model = MagicMock()
+        mock_model.to = MagicMock(return_value=mock_model)
+        mock_model.eval = MagicMock()
+        mock_model.config = MagicMock()
+        mock_model.config.id2label = None
+        mock_proc = MagicMock()
+
+        with (
+            patch(
+                "glmocr.layout.layout_detector.PPDocLayoutV3ForObjectDetection.from_pretrained",
+                return_value=mock_model,
+            ),
+            patch(
+                "glmocr.layout.layout_detector.PPDocLayoutV3ImageProcessorFast.from_pretrained",
+                return_value=mock_proc,
+            ),
+            pytest.raises(RuntimeError, match="Missing id2label"),
+        ):
+            det.start()
+
 
 class TestPageLoader:
     """Tests for PageLoader."""
diff --git a/pyproject.toml b/pyproject.toml
index 9c37852..63e5dc1 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -43,6 +43,7 @@ layout = [
   "transformers>=5.3.0",
   "sentencepiece>=0.2.0",
   "accelerate>=1.13.0",
+  "opencv-python>=4.8.0",
 ]
 
 server = [
@@ -55,6 +56,7 @@ selfhosted = [
   "transformers>=5.3.0",
   "sentencepiece>=0.2.0",
   "accelerate>=1.13.0",
+  "opencv-python>=4.8.0",
   "pypdfium2>=5.6.0",
 ]
 
@@ -64,6 +66,7 @@ all = [
   "transformers>=5.3.0",
   "sentencepiece>=0.2.0",
   "accelerate>=1.13.0",
+  "opencv-python>=4.8.0",
   "pypdfium2>=5.6.0",
   "flask>=3.1.0",
 ]