Better Automodel Design (Layout-Parser#146)

* improve automodel design
1. raise warning for no available models
2. identify models based on dataset names

* more tests for automodels

* fix paddle detection protobuf error?

* Case matters

Author: lolipopshock

.github/workflows/ci.yml

@@ -61,6 +61,8 @@ jobs:
         pip install pytest
         pip install -e ".[paddledetection]"
         pytest tests_deps/test_only_paddledetection.py
+      env:
+        PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION: python
 
   test_all_methods_all_backends:
     needs: [test_only_effdet_backend, test_only_detectron2_backend, test_only_paddledetection_backend]

src/layoutparser/models/auto_layoutmodel.py

@@ -13,26 +13,57 @@
 # limitations under the License.
 
 from typing import Optional, Dict, Union, List
-from .detectron2.layoutmodel import Detectron2LayoutModel
-from .paddledetection.layoutmodel import PaddleDetectionLayoutModel
-from .effdet.layoutmodel import EfficientDetLayoutModel
+from collections import defaultdict
+
 from .model_config import (
     is_lp_layout_model_config_any_format,
 )
+from ..file_utils import (
+    is_effdet_available,
+    is_detectron2_available,
+    is_paddle_available,
+)
+
+ALL_AVAILABLE_BACKENDS = dict()
+ALL_AVAILABLE_DATASETS = defaultdict(list)
+
+if is_effdet_available():
+    from .effdet.layoutmodel import EfficientDetLayoutModel
+    from .effdet.catalog import MODEL_CATALOG as _effdet_model_catalog
+
+    # fmt: off
+    ALL_AVAILABLE_BACKENDS[EfficientDetLayoutModel.DETECTOR_NAME] = EfficientDetLayoutModel
+    for dataset_name in _effdet_model_catalog:
+        ALL_AVAILABLE_DATASETS[dataset_name].append(EfficientDetLayoutModel.DETECTOR_NAME)
+    # fmt: on
+
+if is_detectron2_available():
+    from .detectron2.layoutmodel import Detectron2LayoutModel
+    from .detectron2.catalog import MODEL_CATALOG as _detectron2_model_catalog
+
+    # fmt: off
+    ALL_AVAILABLE_BACKENDS[Detectron2LayoutModel.DETECTOR_NAME] = Detectron2LayoutModel
+    for dataset_name in _detectron2_model_catalog:
+        ALL_AVAILABLE_DATASETS[dataset_name].append(Detectron2LayoutModel.DETECTOR_NAME)
+    # fmt: on
+
+if is_paddle_available():
+    from .paddledetection.layoutmodel import PaddleDetectionLayoutModel
+    from .paddledetection.catalog import MODEL_CATALOG as _paddle_model_catalog
 
-ALL_AVAILABLE_BACKENDS = {
-    Detectron2LayoutModel.DETECTOR_NAME: Detectron2LayoutModel,
-    PaddleDetectionLayoutModel.DETECTOR_NAME: PaddleDetectionLayoutModel,
-    EfficientDetLayoutModel.DETECTOR_NAME: EfficientDetLayoutModel,
-}
+    # fmt: off
+    ALL_AVAILABLE_BACKENDS[PaddleDetectionLayoutModel.DETECTOR_NAME] = PaddleDetectionLayoutModel
+    for dataset_name in _paddle_model_catalog:
+        ALL_AVAILABLE_DATASETS[dataset_name].append(PaddleDetectionLayoutModel.DETECTOR_NAME)
+    # fmt: on
 
 
 def AutoLayoutModel(
     config_path: str,
     model_path: Optional[str] = None,
-    label_map: Optional[Dict]=None,
-    device: Optional[str]=None,
-    extra_config: Optional[Union[Dict, List]]=None,
+    label_map: Optional[Dict] = None,
+    device: Optional[str] = None,
+    extra_config: Optional[Union[Dict, List]] = None,
 ) -> "BaseLayoutModel":
     """[summary]
 
@@ -50,7 +81,7 @@ def AutoLayoutModel(
             Defaults to `None`.
         device(:obj:`str`, optional):
             Whether to use cuda or cpu devices. If not set, LayoutParser will
-            automatically determine the device to initialize the models on. 
+            automatically determine the device to initialize the models on.
         extra_config (:obj:`dict`, optional):
             Extra configuration passed used for initializing the layout model.
 
@@ -59,6 +90,8 @@ def AutoLayoutModel(
     """
     if not is_lp_layout_model_config_any_format(config_path):
         raise ValueError(f"Invalid model config_path {config_path}")
+    
+    # Try to search for the model keywords 
     for backend_name in ALL_AVAILABLE_BACKENDS:
         if backend_name in config_path:
             return ALL_AVAILABLE_BACKENDS[backend_name](
@@ -68,3 +101,16 @@ def AutoLayoutModel(
                 extra_config=extra_config,
                 device=device,
             )
+    
+    # Try to search for the dataset keywords 
+    for dataset_name in ALL_AVAILABLE_DATASETS:
+        if dataset_name in config_path:
+            return ALL_AVAILABLE_BACKENDS[ALL_AVAILABLE_DATASETS[dataset_name][0]](
+                config_path,
+                model_path=model_path,
+                label_map=label_map,
+                extra_config=extra_config,
+                device=device,
+            )
+
+    raise ValueError(f"No available model found for {config_path}")

tests/test_model.py

@@ -44,11 +44,6 @@
     "lp://MFD/tf_efficientdet_d1/config",
 ]
 
-AUTOMODEL_CONFIGS = [
-    "lp://detectron2/PubLayNet/faster_rcnn_R_50_FPN_3x/config",
-    "lp://paddledetection/PubLayNet/ppyolov2_r50vd_dcn_365e/config",
-    "lp://efficientdet/PubLayNet/tf_efficientdet_d0/config",
-]
 
 def _construct_valid_config_variations(config, backend_name):
     dataset_name, arch_name, identifier = config[len("lp://") :].split("/")
@@ -152,8 +147,36 @@ def test_EffDetModel(is_large_scale=False):
             EfficientDetLayoutModel, ALL_EFFDET_MODEL_CONFIGS[0]
         )
 
+
 def test_AutoModel():
-    for config in AUTOMODEL_CONFIGS:
+
+    # Full configs
+    auto_model_config_1 = [
+        "lp://detectron2/PubLayNet/faster_rcnn_R_50_FPN_3x/config",
+        "lp://paddledetection/PubLayNet/ppyolov2_r50vd_dcn_365e/config",
+        "lp://efficientdet/PubLayNet/tf_efficientdet_d0/config",
+    ]
+    for config in auto_model_config_1:
         model = AutoLayoutModel(config)
         image = cv2.imread("tests/fixtures/model/test_model_image.jpg")
-        layout = model.detect(image)
+        layout = model.detect(image)
+
+    # Dataset name only
+    # It will use the first available model
+    auto_model_config_2 = [
+        "lp://PubLayNet",
+        "lp://MFD",
+    ]
+    for config in auto_model_config_1:
+        model = AutoLayoutModel(config)
+        model.DETECTOR_NAME == "efficientdet"
+
+    # Automodel name that doesn't work
+
+    # 1. No available backend for the model
+    with pytest.raises(ValueError):
+        model = AutoLayoutModel("lp://prima")
+
+    # 2. Completely invalid name
+    with pytest.raises(ValueError):
+        model = AutoLayoutModel("lp://test")