add ppv6

.gitignore

@@ -1,2 +1,3 @@
+PPv6/
 out.json
 .vscode/settings.json

.vscode/launch.json

@@ -0,0 +1,12 @@
+{
+    "version": "0.2.0",
+    "configurations": [
+        {
+            "name": "Python Debugger: Current File",
+            "type": "debugpy",
+            "request": "launch",
+            "program": "${file}",
+            "console": "integratedTerminal"
+        }
+    ]
+}

main.py

@@ -1,6 +1,8 @@
 
 # see huggingface/project/flask_auto_selection.py
 
+# see huggingface_echodict/typst-app-clone/tools/extract_pdf_images.py
+
 # C:\Users\echod\.paddlex\official_models\PP-OCRv5_server_det
 # c:\Users\echod\.conda\envs\ppv5\lib\site-packages\paddle\utils\cpp_extension\extension_utils.py 看模型加载的代码在哪
 # PP-OCRv5_server_det PP-OCRv5_server_det.yaml  搜这两个

main_v6.py

@@ -0,0 +1,44 @@
+
+# see huggingface_echodict/typst-app-clone/tools/extract_pdf_images.py
+
+# see https://huggingface.co/PaddlePaddle/PP-OCRv6_medium_det_safetensors
+# see https://huggingface.co/PaddlePaddle/PP-OCRv6_medium_rec_safetensors
+
+"""
+
+curl -LsSf https://astral.sh/uv/install.sh | sh
+	# On Windows.
+powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"
+
+uv venv --python 3.10 --seed --clear
+
+.venv\Scripts\pip.exe install "paddleocr[all]"
+
+.venv\Scripts\pip.exe install transformers torch
+
+.venv\Scripts\pip.exe install paddlepaddle -i https://www.paddlepaddle.org.cn/packages/stable/cpu/
+
+"""
+
+if __name__ == '__main__':
+
+    from paddleocr import PaddleOCR
+
+    ocr = PaddleOCR(
+        text_detection_model_dir="./PPv6/PP-OCRv6_medium_det_safetensors",
+        text_recognition_model_dir="./PPv6/PP-OCRv6_medium_rec_safetensors",
+        lang='chinese_cht',                  # 繁体字典
+        use_doc_orientation_classify=True,   # 整页方向（横/倒）
+        use_doc_unwarping=True,              # 弯曲矫正，扫描古籍有用
+        use_textline_orientation=True,       # 文本行方向分类，竖排靠它
+        text_det_thresh=0.1,                 # 默认 0.3 对古籍太高，漏淡墨
+        text_det_box_thresh=0.1,             # 同上
+        text_rec_score_thresh=0.3,           # 过滤低置信，古籍可放低
+    )
+    result = ocr.predict("./data/SWX0005_00000_00001.webp")
+    for res in result:
+        res.print()
+        res.save_to_img("output")
+        res.save_to_json("output")
+
+    pass