Merge branch 'Dev'

.gitignore

@@ -129,9 +129,9 @@ dmypy.json
 .pyre/
 
 # Custom
+.idea
 Data
 Output
 Models
 MonlamOCR/Models
-
 Map_eText.ipynb

MonlamOCR/Config.py

@@ -0,0 +1,60 @@
+import os
+from huggingface_hub import snapshot_download
+
+
+MODEL_DICT = {
+    "Woodblock": "BDRC/Woodblock",
+    "DergeTenjur": "BDRC/DergeTenjur",
+    "GoogleBooks_C": "BDRC/GoogleBooks_C_v1",
+    "GoogleBooks_E": "BDRC/GoogleBooks_E_v1",
+}
+
+
+# download the line model: https://huggingface.co/BDRC/PhotiLines
+def init_monlam_line_model() -> str:
+    model_id = "BDRC/PhotiLines"
+    model_path = snapshot_download(
+        repo_id=model_id,
+        repo_type="model",
+        local_dir=f"Models/{model_id}",
+    )
+    model_config = f"{model_path}/config.json"
+    assert os.path.isfile(model_config)
+
+    return model_config
+
+
+# download the layout model: https://huggingface.co/BDRC/Photi
+def init_monlam_layout_model() -> str:
+    model_id = "BDRC/Photi"
+    model_path = snapshot_download(
+        repo_id=model_id,
+        repo_type="model",
+        local_dir=f"Models/{model_id}",
+    )
+
+    model_config = f"{model_path}/config.json"
+    assert os.path.isfile(model_config)
+
+    return model_config
+
+
+def init_monla_ocr_model(identifier: str) -> str:
+    available_models = list(MODEL_DICT.keys())
+
+    if identifier in available_models:
+        model_id = MODEL_DICT[identifier]
+
+        model_path = snapshot_download(
+            repo_id=model_id,
+            repo_type="model",
+            local_dir=f"Models/{model_id}",
+        )
+
+        model_config = f"{model_path}/config.json"
+        assert os.path.isfile(model_config)
+
+        return model_config
+    else:
+        print(f"Error: {identifier} is not available")
+        return None

MonlamOCR/Data.py

@@ -16,11 +16,40 @@ class Line:
     bbox: BBox
     center: tuple[int, int]
 
+
+@dataclass
+class LineData:
+    image: npt.NDArray
+    prediction: npt.NDArray
+    angle: float
+    lines: list[Line]
+
+
+@dataclass
+class LayoutData:
+    image: npt.NDArray
+    rotation: float
+    images: list[BBox]
+    text_bboxes: list[BBox]
+    lines: list[Line]
+    captions: list[BBox]
+    margins: list[BBox]
+    predictions: dict[str, npt.NDArray]
+
+
 @dataclass
 class LineDetectionConfig:
     model_file: str
     patch_size: int
 
+
+@dataclass
+class LayoutDetectionConfig:
+    model_file: str
+    patch_size: int
+    classes: list[str]
+
+
 @dataclass
 class OCRConfig:
     model_file: str