deploy: 1a5a172

RapidAI · Feb 28, 2024 · 565f5ca · 565f5ca
1 parent c89aaa9
commit 565f5ca
Show file tree

Hide file tree

Showing 49 changed files with 186 additions and 130 deletions.
diff --git a/docs/about_model/convert_model/index.html b/docs/about_model/convert_model/index.html
@@ -1390,7 +1390,7 @@ <h3 id="离线转换httpsgithubcomrapidaipaddleocrmodelconverter"><a href="https
                 href: "\/RapidOCRDocs\/docs\/install_usage\/rapidocr\/usage\/",
                 title: "使用说明",
                 description: "初始化 link类RapidOCR是主类，其初始化函数如下：\nclass RapidOCR: def __init__(self, config_path: Optional[str] = None, **kwargs): pass 支持两种自定义传参数的方案，下面分别详细说明：\n以config.yaml方式\n找到rapidocr_onnxruntime安装目录下的config.yaml文件，可以通过pip show rapidocr_onnxruntime找到其安装路径。 将config.yaml拷贝出来，放到当前运行目录下 按需自定义参数修改即可，具体参数解释，参见config.yaml engine = RapidOCR(config_path=\"your.yaml\") (推荐) 以具体参数传入，参数基本和config.yaml中对应，只是个别名称有所区别。\ninfo 以下参数均有默认值，可以不传入任何参数，直接初始化使用即可。 class RapidOCR: def __init__( self, text_score: float = 0.5, print_verbose: bool = False, min_height: int = 30, width_height_ratio: float = 8, det_use_cuda: bool = False, det_model_path: Optional[str] = None, det_limit_side_len: float = 736, det_limit_type: str = \"min\", det_thresh: float = 0.3, det_box_thresh: float = 0.",
-                content: "初始化 link类RapidOCR是主类，其初始化函数如下：\nclass RapidOCR: def __init__(self, config_path: Optional[str] = None, **kwargs): pass 支持两种自定义传参数的方案，下面分别详细说明：\n以config.yaml方式\n找到rapidocr_onnxruntime安装目录下的config.yaml文件，可以通过pip show rapidocr_onnxruntime找到其安装路径。 将config.yaml拷贝出来，放到当前运行目录下 按需自定义参数修改即可，具体参数解释，参见config.yaml engine = RapidOCR(config_path=\"your.yaml\") (推荐) 以具体参数传入，参数基本和config.yaml中对应，只是个别名称有所区别。\ninfo 以下参数均有默认值，可以不传入任何参数，直接初始化使用即可。 class RapidOCR: def __init__( self, text_score: float = 0.5, print_verbose: bool = False, min_height: int = 30, width_height_ratio: float = 8, det_use_cuda: bool = False, det_model_path: Optional[str] = None, det_limit_side_len: float = 736, det_limit_type: str = \"min\", det_thresh: float = 0.3, det_box_thresh: float = 0.5, det_unclip_ratio: float = 1.6, det_donot_use_dilation: bool = False, det_score_mode: str = \"fast\", cls_use_cuda: bool = False, cls_model_path: Optional[str] = None, cls_image_shape: List[int] = [3, 48, 192], cls_label_list: List[str] = [\"0\", \"180\"], cls_batch_num: int = 6, cls_thresh: float = 0.9, rec_use_cuda: bool = False, rec_model_path: Optional[str] = None, rec_img_shape: List[int] = [3, 48, 320], rec_batch_num: int = 6, ): pass engine = RapidOCR() res, elapse = engine(img, use_det=True, use_cls=True, use_rec=True) 输入 link支持4种输入类型：Union[str, np.ndarray, bytes, Path]\nstr np.ndarray Bytes Path from rapidocr_onnxruntime import RapidOCR engine = RapidOCR() img_path = 'tests/test_files/ch_en_num.jpg' result, elapse = engine(img_path) print(result) print(elapse) import cv2 from rapidocr_onnxruntime import RapidOCR engine = RapidOCR() img = cv2.imread('tests/test_files/ch_en_num.jpg') result, elapse = engine(img) print(result) print(elapse) from rapidocr_onnxruntime import RapidOCR engine = RapidOCR() img_path = 'tests/test_files/ch_en_num.jpg' with open(img_path, 'rb') as f: img = f.read() result, elapse = engine(img) print(result) print(elapse) from pathlib import Path from rapidocr_onnxruntime import RapidOCR engine = RapidOCR() img_path = Path('tests/test_files/ch_en_num.jpg') result, elapse = engine(img_path) print(result) print(elapse) 输出 link类RapidOCR在调用时，有三个参数use_det | use_cls | use_rec，可以控制是否使用检测、方向分类和识别这三部分。不同的参数，决定了不同的输出，详细搭配如下：\n只有检测 只有分类 只有识别 检测 + 识别 分类 + 识别 检测 + 分类 + 识别 from rapidocr_onnxruntime import RapidOCR engine = RapidOCR() img_path = 'tests/test_files/ch_en_num.jpg' result, elapse = engine(img_path, use_det=True, use_cls=False, use_rec=False) print(result) print(elapse) 返回值result: List[List[float]] (每个框的坐标[左上, 右上, 右下, 左下])\n[ [[5.0, 2.0], [322.0, 9.0], [319.0, 103.0], [3.0, 96.0]], [[70.0, 98.0], [252.0, 97.0], [252.0, 125.0], [70.0, 126.0]], ... ] from rapidocr_onnxruntime import RapidOCR engine = RapidOCR() img_path = 'tests/test_files/ch_en_num.jpg' result, elapse = engine(img_path, use_det=False, use_cls=True, use_rec=False) print(result) print(elapse) 返回值result: List[List[str, float]] ([方向0或180, 置信度])\n[ ['0', 0.9998784], ... ] from rapidocr_onnxruntime import RapidOCR engine = RapidOCR() img_path = 'tests/test_files/ch_en_num.jpg' result, elapse = engine(img_path, use_det=False, use_cls=False, use_rec=True) print(result) print(elapse) 返回值result: List[List[str, float]] ([识别的文本, 置信度])\n[ ['韩国小馆', 0.7992169380187988], ... ] from rapidocr_onnxruntime import RapidOCR engine = RapidOCR() img_path = 'tests/test_files/ch_en_num.jpg' result, elapse = engine(img_path, use_det=True, use_cls=False, use_rec=True) print(result) print(elapse) 返回值result: List[List[float], str, float] ([[左上, 右上, 右下, 左下], 文本内容, 置信度])\n[ [[[9.0, 2.0], [321.0, 11.0], [318.0, 102.0], [6.0, 93.0]], '正品促销', '0.7986101984977723'], [[[70.0, 98.0], [251.0, 98.0], [251.0, 125.0], [70.0, 125.0]], '大桶装更划算', '0.7368737288883754'], ... ] from rapidocr_onnxruntime import RapidOCR engine = RapidOCR() img_path = 'tests/test_files/ch_en_num.jpg' result, elapse = engine(img_path, use_det=False, use_cls=True, use_rec=True) print(result) print(elapse) 返回值result: List[List[str, float]] ([识别的文本, 置信度])\n[ ['韩国小馆', 0.7992169380187988], ... ] from rapidocr_onnxruntime import RapidOCR engine = RapidOCR() img_path = 'tests/test_files/ch_en_num.jpg' # 默认都为True result, elapse = engine(img_path, use_det=True, use_cls=True, use_rec=True) print(result) print(elapse) 返回值result: List[List[float], str, float] ([[左上, 右上, 右下, 左下], 文本内容, 置信度])\n[ [[[9.0, 2.0], [321.0, 11.0], [318.0, 102.0], [6.0, 93.0]], '正品促销', '0.7986101984977723'], [[[70.0, 98.0], [251.0, 98.0], [251.0, 125.0], [70.0, 125.0]], '大桶装更划算', '0.7368737288883754'], ... ] 可视化查看结果 link为了便于查看检测和识别结果，该库中封装了VisRes类，可借助该类快速可视化查看结果。\ninfo 可视化时，需要提供字体文件。下载链接：link\n只可视化检测 可视化检测和识别 import cv2 from rapidocr_onnxruntime import RapidOCR, VisRes engine = RapidOCR() vis = VisRes() image_path = \"tests/test_files/ch_en_num.jpg\" img = cv2.imread(image_path) result, elapse_list = engine(img) boxes, txts, scores = list(zip(*result)) res = vis(img, boxes) cv2.imwrite(\"only_vis_det.png\", res) import cv2 from rapidocr_onnxruntime import RapidOCR, VisRes engine = RapidOCR() vis = VisRes() image_path = \"tests/test_files/ch_en_num.jpg\" img = cv2.imread(image_path) result, elapse_list = engine(img) boxes, txts, scores = list(zip(*result)) font_path=\"resources/fonts/FZYTK.TTF\" res = vis(img, boxes, txts, scores, font_path) cv2.imwrite(\"vis_det_rec.png\", res) "
+                content: "初始化 link类RapidOCR是主类，其初始化函数如下：\nclass RapidOCR: def __init__(self, config_path: Optional[str] = None, **kwargs): pass 支持两种自定义传参数的方案，下面分别详细说明：\n以config.yaml方式\n找到rapidocr_onnxruntime安装目录下的config.yaml文件，可以通过pip show rapidocr_onnxruntime找到其安装路径。 将config.yaml拷贝出来，放到当前运行目录下 按需自定义参数修改即可，具体参数解释，参见config.yaml engine = RapidOCR(config_path=\"your.yaml\") (推荐) 以具体参数传入，参数基本和config.yaml中对应，只是个别名称有所区别。\ninfo 以下参数均有默认值，可以不传入任何参数，直接初始化使用即可。 class RapidOCR: def __init__( self, text_score: float = 0.5, print_verbose: bool = False, min_height: int = 30, width_height_ratio: float = 8, det_use_cuda: bool = False, det_model_path: Optional[str] = None, det_limit_side_len: float = 736, det_limit_type: str = \"min\", det_thresh: float = 0.3, det_box_thresh: float = 0.5, det_unclip_ratio: float = 1.6, det_donot_use_dilation: bool = False, det_score_mode: str = \"fast\", cls_use_cuda: bool = False, cls_model_path: Optional[str] = None, cls_image_shape: List[int] = [3, 48, 192], cls_label_list: List[str] = [\"0\", \"180\"], cls_batch_num: int = 6, cls_thresh: float = 0.9, rec_use_cuda: bool = False, rec_model_path: Optional[str] = None, rec_img_shape: List[int] = [3, 48, 320], rec_batch_num: int = 6, ): pass engine = RapidOCR() res, elapse = engine(img, use_det=True, use_cls=True, use_rec=True) 输入 link info 确保输入模型前的图像通道顺序为BGR即可。当前LoadImage类内部已经对此做了处理，参考下面写法即可正常使用。 支持4种输入类型：Union[str, np.ndarray, bytes, Path, PIL.Image.Image]\nstr np.ndarray PIL.Image.Image Bytes Path from rapidocr_onnxruntime import RapidOCR engine = RapidOCR() img_path = 'tests/test_files/ch_en_num.jpg' result, elapse = engine(img_path) print(result) print(elapse) import cv2 from rapidocr_onnxruntime import RapidOCR engine = RapidOCR() img = cv2.imread('tests/test_files/ch_en_num.jpg') result, elapse = engine(img) print(result) print(elapse) from PIL import Image from rapidocr_onnxruntime import RapidOCR engine = RapidOCR() img = Image.open('tests/test_files/ch_en_num.jpg') result, elapse = engine(img) print(result) print(elapse) from rapidocr_onnxruntime import RapidOCR engine = RapidOCR() img_path = 'tests/test_files/ch_en_num.jpg' with open(img_path, 'rb') as f: img = f.read() result, elapse = engine(img) print(result) print(elapse) from pathlib import Path from rapidocr_onnxruntime import RapidOCR engine = RapidOCR() img_path = Path('tests/test_files/ch_en_num.jpg') result, elapse = engine(img_path) print(result) print(elapse) 输出 link类RapidOCR在调用时，有三个参数use_det | use_cls | use_rec，可以控制是否使用检测、方向分类和识别这三部分。不同的参数，决定了不同的输出，详细搭配如下：\n只有检测 只有分类 只有识别 检测 + 识别 分类 + 识别 检测 + 分类 + 识别 from rapidocr_onnxruntime import RapidOCR engine = RapidOCR() img_path = 'tests/test_files/ch_en_num.jpg' result, elapse = engine(img_path, use_det=True, use_cls=False, use_rec=False) print(result) print(elapse) 返回值result: List[List[float]] (每个框的坐标[左上, 右上, 右下, 左下])\n[ [[5.0, 2.0], [322.0, 9.0], [319.0, 103.0], [3.0, 96.0]], [[70.0, 98.0], [252.0, 97.0], [252.0, 125.0], [70.0, 126.0]], ... ] from rapidocr_onnxruntime import RapidOCR engine = RapidOCR() img_path = 'tests/test_files/ch_en_num.jpg' result, elapse = engine(img_path, use_det=False, use_cls=True, use_rec=False) print(result) print(elapse) 返回值result: List[List[str, float]] ([方向0或180, 置信度])\n[ ['0', 0.9998784], ... ] from rapidocr_onnxruntime import RapidOCR engine = RapidOCR() img_path = 'tests/test_files/ch_en_num.jpg' result, elapse = engine(img_path, use_det=False, use_cls=False, use_rec=True) print(result) print(elapse) 返回值result: List[List[str, float]] ([识别的文本, 置信度])\n[ ['韩国小馆', 0.7992169380187988], ... ] from rapidocr_onnxruntime import RapidOCR engine = RapidOCR() img_path = 'tests/test_files/ch_en_num.jpg' result, elapse = engine(img_path, use_det=True, use_cls=False, use_rec=True) print(result) print(elapse) 返回值result: List[List[float], str, float] ([[左上, 右上, 右下, 左下], 文本内容, 置信度])\n[ [[[9.0, 2.0], [321.0, 11.0], [318.0, 102.0], [6.0, 93.0]], '正品促销', '0.7986101984977723'], [[[70.0, 98.0], [251.0, 98.0], [251.0, 125.0], [70.0, 125.0]], '大桶装更划算', '0.7368737288883754'], ... ] from rapidocr_onnxruntime import RapidOCR engine = RapidOCR() img_path = 'tests/test_files/ch_en_num.jpg' result, elapse = engine(img_path, use_det=False, use_cls=True, use_rec=True) print(result) print(elapse) 返回值result: List[List[str, float]] ([识别的文本, 置信度])\n[ ['韩国小馆', 0.7992169380187988], ... ] from rapidocr_onnxruntime import RapidOCR engine = RapidOCR() img_path = 'tests/test_files/ch_en_num.jpg' # 默认都为True result, elapse = engine(img_path, use_det=True, use_cls=True, use_rec=True) print(result) print(elapse) 返回值result: List[List[float], str, float] ([[左上, 右上, 右下, 左下], 文本内容, 置信度])\n[ [[[9.0, 2.0], [321.0, 11.0], [318.0, 102.0], [6.0, 93.0]], '正品促销', '0.7986101984977723'], [[[70.0, 98.0], [251.0, 98.0], [251.0, 125.0], [70.0, 125.0]], '大桶装更划算', '0.7368737288883754'], ... ] 可视化查看结果 link为了便于查看检测和识别结果，该库中封装了VisRes类，可借助该类快速可视化查看结果。\ninfo 可视化时，需要提供字体文件。下载链接：link\n只可视化检测 可视化检测和识别 import cv2 from rapidocr_onnxruntime import RapidOCR, VisRes engine = RapidOCR() vis = VisRes() image_path = \"tests/test_files/ch_en_num.jpg\" img = cv2.imread(image_path) result, elapse_list = engine(img) boxes, txts, scores = list(zip(*result)) res = vis(img, boxes) cv2.imwrite(\"only_vis_det.png\", res) import cv2 from rapidocr_onnxruntime import RapidOCR, VisRes engine = RapidOCR() vis = VisRes() image_path = \"tests/test_files/ch_en_num.jpg\" img = cv2.imread(image_path) result, elapse_list = engine(img) boxes, txts, scores = list(zip(*result)) font_path=\"resources/fonts/FZYTK.TTF\" res = vis(img, boxes, txts, scores, font_path) cv2.imwrite(\"vis_det_rec.png\", res) "
             }
         );
     index.add(