Add Hub download metadata

README.md

@@ -17,7 +17,7 @@ metrics:
 - precision
 - recall
 - f1
-library_name: transformers
+library_name: onnxruntime
 ---
 
 # Hiro-Layout: Document Layout Analysis for Patent and Technical PDFs
@@ -153,12 +153,18 @@ The current model artifact is an ONNX export:
 layout_model/RT-DETR_25.onnx
 ```
 
-The model can be loaded with ONNXRuntime:
+Download the repository from the Hugging Face Hub and load the ONNX model with ONNXRuntime:
 
 ```python
+from pathlib import Path
+
+from huggingface_hub import snapshot_download
 import onnxruntime as ort
 
-session = ort.InferenceSession("layout_model/RT-DETR_25.onnx")
+repo_dir = snapshot_download("PatSnap/Hiro-Layout")
+model_path = Path(repo_dir) / "layout_model" / "RT-DETR_25.onnx"
+
+session = ort.InferenceSession(str(model_path))
 print("inputs:", [i.name for i in session.get_inputs()])
 print("outputs:", [o.name for o in session.get_outputs()])
 ```
@@ -171,6 +177,7 @@ Use `labels.json` for the 25-class label mapping.
 | --- | --- |
 | `README.md` | Hugging Face model card in English |
 | `README_zh.md` | Chinese model card |
+| `config.json` | Model metadata used by Hugging Face Hub tooling and download statistics |
 | `EVALUATION.md` | Detailed benchmark results derived from the workbook |
 | `labels.json` | Machine-readable 25-class label mapping |
 | `layout_model/RT-DETR_25.onnx` | ONNX model artifact |

README_zh.md

@@ -17,7 +17,7 @@ metrics:
 - precision
 - recall
 - f1
-library_name: transformers
+library_name: onnxruntime
 ---
 
 # Hiro-Layout：面向专利和技术 PDF 的文档版面分析模型
@@ -153,12 +153,18 @@ Hiro-Layout 是一个面向专利和技术 PDF 页面图像的文档版面分析
 layout_model/RT-DETR_25.onnx
 ```
 
-模型可使用 ONNXRuntime 加载：
+先从 Hugging Face Hub 下载仓库，再使用 ONNXRuntime 加载模型：
 
 ```python
+from pathlib import Path
+
+from huggingface_hub import snapshot_download
 import onnxruntime as ort
 
-session = ort.InferenceSession("layout_model/RT-DETR_25.onnx")
+repo_dir = snapshot_download("PatSnap/Hiro-Layout")
+model_path = Path(repo_dir) / "layout_model" / "RT-DETR_25.onnx"
+
+session = ort.InferenceSession(str(model_path))
 print("inputs:", [i.name for i in session.get_inputs()])
 print("outputs:", [o.name for o in session.get_outputs()])
 ```
@@ -171,6 +177,7 @@ print("outputs:", [o.name for o in session.get_outputs()])
 | --- | --- |
 | `README.md` | 英文 Hugging Face 模型卡 |
 | `README_zh.md` | 中文模型卡 |
+| `config.json` | Hugging Face Hub 工具和下载统计使用的模型元数据 |
 | `EVALUATION.md` | 从 Excel 整理出的详细评测结果 |
 | `labels.json` | 机器可读的 25 类标签映射 |
 | `layout_model/RT-DETR_25.onnx` | ONNX 模型文件 |

config.json

@@ -0,0 +1,65 @@
+{
+  "model_type": "hiro-layout",
+  "architectures": [
+    "RT-DETR"
+  ],
+  "task": "object-detection",
+  "library_name": "onnxruntime",
+  "model_file": "layout_model/RT-DETR_25.onnx",
+  "label_file": "labels.json",
+  "num_labels": 25,
+  "id2label": {
+    "0": "graph",
+    "1": "drawing",
+    "2": "structure diagram",
+    "3": "photograph",
+    "4": "table",
+    "5": "math equation",
+    "6": "chemical formula",
+    "7": "noise",
+    "8": "text",
+    "9": "title",
+    "10": "section title",
+    "11": "page header",
+    "12": "page footer",
+    "13": "marginal note",
+    "14": "caption",
+    "15": "figure number",
+    "16": "line number",
+    "17": "column number",
+    "18": "sequence",
+    "19": "figure complex",
+    "20": "chemical reaction",
+    "21": "bibliography",
+    "22": "search report",
+    "23": "Table of Contents",
+    "24": "reference"
+  },
+  "label2id": {
+    "graph": 0,
+    "drawing": 1,
+    "structure diagram": 2,
+    "photograph": 3,
+    "table": 4,
+    "math equation": 5,
+    "chemical formula": 6,
+    "noise": 7,
+    "text": 8,
+    "title": 9,
+    "section title": 10,
+    "page header": 11,
+    "page footer": 12,
+    "marginal note": 13,
+    "caption": 14,
+    "figure number": 15,
+    "line number": 16,
+    "column number": 17,
+    "sequence": 18,
+    "figure complex": 19,
+    "chemical reaction": 20,
+    "bibliography": 21,
+    "search report": 22,
+    "Table of Contents": 23,
+    "reference": 24
+  }
+}