onnxtr 0.4.0__tar.gz → 0.5.0__tar.gz

{onnxtr-0.4.0 → onnxtr-0.5.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: onnxtr
-Version: 0.4.0
+Version: 0.5.0
 Summary: Onnx Text Recognition (OnnxTR): docTR Onnx-Wrapper for high-performance OCR on documents.
 Author-email: Felix Dittrich <felixdittrich92@gmail.com>
 Maintainer: Felix Dittrich
@@ -225,9 +225,8 @@ Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
 Requires-Python: <4,>=3.9.0
 Description-Content-Type: text/markdown
 License-File: LICENSE
-Requires-Dist: numpy<2.0.0,>=1.16.0
+Requires-Dist: numpy<3.0.0,>=1.16.0
 Requires-Dist: scipy<2.0.0,>=1.4.0
-Requires-Dist: opencv-python<5.0.0,>=4.5.0
 Requires-Dist: pypdfium2<5.0.0,>=4.11.0
 Requires-Dist: pyclipper<2.0.0,>=1.2.0
 Requires-Dist: shapely<3.0.0,>=1.6.0
@@ -240,8 +239,16 @@ Requires-Dist: anyascii>=0.3.2
 Requires-Dist: tqdm>=4.30.0
 Provides-Extra: cpu
 Requires-Dist: onnxruntime>=1.11.0; extra == "cpu"
+Requires-Dist: opencv-python<5.0.0,>=4.5.0; extra == "cpu"
 Provides-Extra: gpu
 Requires-Dist: onnxruntime-gpu>=1.11.0; extra == "gpu"
+Requires-Dist: opencv-python<5.0.0,>=4.5.0; extra == "gpu"
+Provides-Extra: cpu-headless
+Requires-Dist: onnxruntime>=1.11.0; extra == "cpu-headless"
+Requires-Dist: opencv-python-headless<5.0.0,>=4.5.0; extra == "cpu-headless"
+Provides-Extra: gpu-headless
+Requires-Dist: onnxruntime-gpu>=1.11.0; extra == "gpu-headless"
+Requires-Dist: opencv-python-headless<5.0.0,>=4.5.0; extra == "gpu-headless"
 Provides-Extra: html
 Requires-Dist: weasyprint>=55.0; extra == "html"
 Provides-Extra: viz
@@ -257,6 +264,7 @@ Requires-Dist: mypy>=0.812; extra == "quality"
 Requires-Dist: pre-commit>=2.17.0; extra == "quality"
 Provides-Extra: dev
 Requires-Dist: onnxruntime>=1.11.0; extra == "dev"
+Requires-Dist: opencv-python<5.0.0,>=4.5.0; extra == "dev"
 Requires-Dist: weasyprint>=55.0; extra == "dev"
 Requires-Dist: matplotlib>=3.1.0; extra == "dev"
 Requires-Dist: mplcursors>=0.3; extra == "dev"
@@ -276,7 +284,7 @@ Requires-Dist: pre-commit>=2.17.0; extra == "dev"
 [![codecov](https://codecov.io/gh/felixdittrich92/OnnxTR/graph/badge.svg?token=WVFRCQBOLI)](https://codecov.io/gh/felixdittrich92/OnnxTR)
 [![Codacy Badge](https://app.codacy.com/project/badge/Grade/4fff4d764bb14fb8b4f4afeb9587231b)](https://app.codacy.com/gh/felixdittrich92/OnnxTR/dashboard?utm_source=gh&utm_medium=referral&utm_content=&utm_campaign=Badge_grade)
 [![CodeFactor](https://www.codefactor.io/repository/github/felixdittrich92/onnxtr/badge)](https://www.codefactor.io/repository/github/felixdittrich92/onnxtr)
-[![Pypi](https://img.shields.io/badge/pypi-v0.3.2-blue.svg)](https://pypi.org/project/OnnxTR/)
+[![Pypi](https://img.shields.io/badge/pypi-v0.5.0-blue.svg)](https://pypi.org/project/OnnxTR/)
 
 > :warning: Please note that this is a wrapper around the [doctr](https://github.com/mindee/doctr) library to provide a Onnx pipeline for docTR. For feature requests, which are not directly related to the Onnx pipeline, please refer to the base project.
 
@@ -309,8 +317,10 @@ For GPU support please take a look at: [ONNX Runtime](https://onnxruntime.ai/get
 
 ```shell
 pip install "onnxtr[cpu]"
+pip install "onnxtr[cpu-headless]"  # same as cpu but with opencv-headless
 # with gpu support
 pip install "onnxtr[gpu]"
+pip install "onnxtr[gpu-headless]"  # same as gpu but with opencv-headless
 # with HTML support
 pip install "onnxtr[html]"
 # with support for visualization
@@ -356,6 +366,9 @@ model = ocr_predictor(
     # Additional parameters - meta information
     detect_orientation=False,  # set to `True` if the orientation of the pages should be detected (default: False)
     detect_language=False, # set to `True` if the language of the pages should be detected (default: False)
+    # Orientation specific parameters in combination with `assume_straight_pages=False` and/or `straighten_pages=True`
+    disable_crop_orientation=False,  # set to `True` if the crop orientation classification should be disabled (default: False)
+    disable_page_orientation=False,  # set to `True` if the general page orientation classification should be disabled (default: False)
     # DocumentBuilder specific parameters
     resolve_lines=True,  # whether words should be automatically grouped into lines (default: True)
     resolve_blocks=False,  # whether lines should be automatically grouped into blocks (default: False)
@@ -589,8 +602,8 @@ The smallest combination in OnnxTR (docTR) of `db_mobilenet_v3_large` and `crnn_
 |Library                          |FUNSD (199 pages)              |CORD  (900 pages)              |
 |---------------------------------|-------------------------------|-------------------------------|
 |docTR (CPU) - v0.8.1             | ~1.29s / Page                 | ~0.60s / Page                 |
-|**OnnxTR (CPU)** - v0.1.2        | ~0.57s / Page                 | **~0.25s / Page**             |
-|**OnnxTR (CPU) 8-bit** - v0.1.2  | **~0.38s / Page**             | **~0.14s / Page**             |
+|**OnnxTR (CPU)** - v0.4.1        | ~0.57s / Page                 | **~0.25s / Page**             |
+|**OnnxTR (CPU) 8-bit** - v0.4.1  | **~0.38s / Page**             | **~0.14s / Page**             |
 |EasyOCR (CPU) - v1.7.1           | ~1.96s / Page                 | ~1.75s / Page                 |
 |**PyTesseract (CPU)** - v0.3.10  | **~0.50s / Page**             | ~0.52s / Page                 |
 |Surya (line) (CPU) - v0.4.4      | ~48.76s / Page                | ~35.49s / Page                |
@@ -602,7 +615,7 @@ The smallest combination in OnnxTR (docTR) of `db_mobilenet_v3_large` and `crnn_
 |-------------------------------------|-------------------------------|-------------------------------|
 |docTR (GPU) - v0.8.1                 | ~0.07s / Page                 | ~0.05s / Page                 |
 |**docTR (GPU) float16** - v0.8.1     | **~0.06s / Page**             | **~0.03s / Page**             |
-|OnnxTR (GPU) - v0.1.2                | **~0.06s / Page**             | ~0.04s / Page                 |
+|OnnxTR (GPU) - v0.4.1                | **~0.06s / Page**             | ~0.04s / Page                 |
 |EasyOCR (GPU) - v1.7.1               | ~0.31s / Page                 | ~0.19s / Page                 |
 |Surya (GPU) float16 - v0.4.4         | ~3.70s / Page                 | ~2.81s / Page                 |
 |**PaddleOCR (GPU) - no cls - v2.7.3**| ~0.08s / Page                 | **~0.03s / Page**             |
@@ -621,6 +634,16 @@ If you wish to cite please refer to the base project citation, feel free to use
 }
 ```
 
+```bibtex
+@misc{onnxtr2024,
+    title={OnnxTR: Optical Character Recognition made seamless & accessible to anyone, powered by Onnx},
+    author={Felix Dittrich},
+    year={2024},
+    publisher = {GitHub},
+    howpublished = {\url{https://github.com/felixdittrich92/OnnxTR}}
+}
+```
+
 ## License
 
 Distributed under the Apache 2.0 License. See [`LICENSE`](https://github.com/felixdittrich92/OnnxTR?tab=Apache-2.0-1-ov-file#readme) for more information.

{onnxtr-0.4.0 → onnxtr-0.5.0}/README.md

@@ -7,7 +7,7 @@
 [![codecov](https://codecov.io/gh/felixdittrich92/OnnxTR/graph/badge.svg?token=WVFRCQBOLI)](https://codecov.io/gh/felixdittrich92/OnnxTR)
 [![Codacy Badge](https://app.codacy.com/project/badge/Grade/4fff4d764bb14fb8b4f4afeb9587231b)](https://app.codacy.com/gh/felixdittrich92/OnnxTR/dashboard?utm_source=gh&utm_medium=referral&utm_content=&utm_campaign=Badge_grade)
 [![CodeFactor](https://www.codefactor.io/repository/github/felixdittrich92/onnxtr/badge)](https://www.codefactor.io/repository/github/felixdittrich92/onnxtr)
-[![Pypi](https://img.shields.io/badge/pypi-v0.3.2-blue.svg)](https://pypi.org/project/OnnxTR/)
+[![Pypi](https://img.shields.io/badge/pypi-v0.5.0-blue.svg)](https://pypi.org/project/OnnxTR/)
 
 > :warning: Please note that this is a wrapper around the [doctr](https://github.com/mindee/doctr) library to provide a Onnx pipeline for docTR. For feature requests, which are not directly related to the Onnx pipeline, please refer to the base project.
 
@@ -40,8 +40,10 @@ For GPU support please take a look at: [ONNX Runtime](https://onnxruntime.ai/get
 
 ```shell
 pip install "onnxtr[cpu]"
+pip install "onnxtr[cpu-headless]"  # same as cpu but with opencv-headless
 # with gpu support
 pip install "onnxtr[gpu]"
+pip install "onnxtr[gpu-headless]"  # same as gpu but with opencv-headless
 # with HTML support
 pip install "onnxtr[html]"
 # with support for visualization
@@ -87,6 +89,9 @@ model = ocr_predictor(
     # Additional parameters - meta information
     detect_orientation=False,  # set to `True` if the orientation of the pages should be detected (default: False)
     detect_language=False, # set to `True` if the language of the pages should be detected (default: False)
+    # Orientation specific parameters in combination with `assume_straight_pages=False` and/or `straighten_pages=True`
+    disable_crop_orientation=False,  # set to `True` if the crop orientation classification should be disabled (default: False)
+    disable_page_orientation=False,  # set to `True` if the general page orientation classification should be disabled (default: False)
     # DocumentBuilder specific parameters
     resolve_lines=True,  # whether words should be automatically grouped into lines (default: True)
     resolve_blocks=False,  # whether lines should be automatically grouped into blocks (default: False)
@@ -320,8 +325,8 @@ The smallest combination in OnnxTR (docTR) of `db_mobilenet_v3_large` and `crnn_
 |Library                          |FUNSD (199 pages)              |CORD  (900 pages)              |
 |---------------------------------|-------------------------------|-------------------------------|
 |docTR (CPU) - v0.8.1             | ~1.29s / Page                 | ~0.60s / Page                 |
-|**OnnxTR (CPU)** - v0.1.2        | ~0.57s / Page                 | **~0.25s / Page**             |
-|**OnnxTR (CPU) 8-bit** - v0.1.2  | **~0.38s / Page**             | **~0.14s / Page**             |
+|**OnnxTR (CPU)** - v0.4.1        | ~0.57s / Page                 | **~0.25s / Page**             |
+|**OnnxTR (CPU) 8-bit** - v0.4.1  | **~0.38s / Page**             | **~0.14s / Page**             |
 |EasyOCR (CPU) - v1.7.1           | ~1.96s / Page                 | ~1.75s / Page                 |
 |**PyTesseract (CPU)** - v0.3.10  | **~0.50s / Page**             | ~0.52s / Page                 |
 |Surya (line) (CPU) - v0.4.4      | ~48.76s / Page                | ~35.49s / Page                |
@@ -333,7 +338,7 @@ The smallest combination in OnnxTR (docTR) of `db_mobilenet_v3_large` and `crnn_
 |-------------------------------------|-------------------------------|-------------------------------|
 |docTR (GPU) - v0.8.1                 | ~0.07s / Page                 | ~0.05s / Page                 |
 |**docTR (GPU) float16** - v0.8.1     | **~0.06s / Page**             | **~0.03s / Page**             |
-|OnnxTR (GPU) - v0.1.2                | **~0.06s / Page**             | ~0.04s / Page                 |
+|OnnxTR (GPU) - v0.4.1                | **~0.06s / Page**             | ~0.04s / Page                 |
 |EasyOCR (GPU) - v1.7.1               | ~0.31s / Page                 | ~0.19s / Page                 |
 |Surya (GPU) float16 - v0.4.4         | ~3.70s / Page                 | ~2.81s / Page                 |
 |**PaddleOCR (GPU) - no cls - v2.7.3**| ~0.08s / Page                 | **~0.03s / Page**             |
@@ -352,6 +357,16 @@ If you wish to cite please refer to the base project citation, feel free to use
 }
 ```
 
+```bibtex
+@misc{onnxtr2024,
+    title={OnnxTR: Optical Character Recognition made seamless & accessible to anyone, powered by Onnx},
+    author={Felix Dittrich},
+    year={2024},
+    publisher = {GitHub},
+    howpublished = {\url{https://github.com/felixdittrich92/OnnxTR}}
+}
+```
+
 ## License
 
 Distributed under the Apache 2.0 License. See [`LICENSE`](https://github.com/felixdittrich92/OnnxTR?tab=Apache-2.0-1-ov-file#readme) for more information.

{onnxtr-0.4.0 → onnxtr-0.5.0}/onnxtr/models/builder.py

@@ -266,7 +266,7 @@ class DocumentBuilder(NestedObject):
                 Line([
                     Word(
                         *word_preds[idx],
-                        tuple([tuple(pt) for pt in boxes[idx].tolist()]),  # type: ignore[arg-type]
+                        tuple(tuple(pt) for pt in boxes[idx].tolist()),  # type: ignore[arg-type]
                         float(objectness_scores[idx]),
                         crop_orientations[idx],
                     )

{onnxtr-0.4.0 → onnxtr-0.5.0}/onnxtr/models/classification/models/mobilenet.py

@@ -13,6 +13,7 @@ import numpy as np
 from ...engine import Engine, EngineConfig
 
 __all__ = [
+    "MobileNetV3",
     "mobilenet_v3_small_crop_orientation",
     "mobilenet_v3_small_page_orientation",
 ]

{onnxtr-0.4.0 → onnxtr-0.5.0}/onnxtr/models/classification/predictor/base.py

@@ -3,7 +3,7 @@
 # This program is licensed under the Apache License 2.0.
 # See LICENSE or go to <https://opensource.org/licenses/Apache-2.0> for full license details.
 
-from typing import Any, List, Union
+from typing import Any, List, Optional, Union
 
 import numpy as np
 from scipy.special import softmax
@@ -29,10 +29,10 @@ class OrientationPredictor(NestedObject):
 
     def __init__(
         self,
-        pre_processor: PreProcessor,
-        model: Any,
+        pre_processor: Optional[PreProcessor],
+        model: Optional[Any],
     ) -> None:
-        self.pre_processor = pre_processor
+        self.pre_processor = pre_processor if isinstance(pre_processor, PreProcessor) else None
         self.model = model
 
     def __call__(
@@ -43,6 +43,10 @@ class OrientationPredictor(NestedObject):
         if any(input.ndim != 3 for input in inputs):
             raise ValueError("incorrect input shape: all inputs are expected to be multi-channel 2D images.")
 
+        if self.model is None or self.pre_processor is None:
+            # predictor is disabled
+            return [[0] * len(inputs), [0] * len(inputs), [1.0] * len(inputs)]
+
         processed_batches = self.pre_processor(inputs)
         predicted_batches = [self.model(batch) for batch in processed_batches]
 

{onnxtr-0.4.0 → onnxtr-0.5.0}/onnxtr/models/classification/zoo.py

@@ -17,16 +17,30 @@ ORIENTATION_ARCHS: List[str] = ["mobilenet_v3_small_crop_orientation", "mobilene
 
 
 def _orientation_predictor(
-    arch: str, load_in_8_bit: bool = False, engine_cfg: Optional[EngineConfig] = None, **kwargs: Any
+    arch: Any,
+    model_type: str,
+    load_in_8_bit: bool = False,
+    engine_cfg: Optional[EngineConfig] = None,
+    disabled: bool = False,
+    **kwargs: Any,
 ) -> OrientationPredictor:
-    if arch not in ORIENTATION_ARCHS:
-        raise ValueError(f"unknown architecture '{arch}'")
+    if disabled:
+        # Case where the orientation predictor is disabled
+        return OrientationPredictor(None, None)
+
+    if isinstance(arch, str):
+        if arch not in ORIENTATION_ARCHS:
+            raise ValueError(f"unknown architecture '{arch}'")
+        # Load directly classifier from backbone
+        _model = classification.__dict__[arch](load_in_8_bit=load_in_8_bit, engine_cfg=engine_cfg)
+    else:
+        if not isinstance(arch, classification.MobileNetV3):
+            raise ValueError(f"unknown architecture: {type(arch)}")
+        _model = arch
 
-    # Load directly classifier from backbone
-    _model = classification.__dict__[arch](load_in_8_bit=load_in_8_bit, engine_cfg=engine_cfg)
     kwargs["mean"] = kwargs.get("mean", _model.cfg["mean"])
     kwargs["std"] = kwargs.get("std", _model.cfg["std"])
-    kwargs["batch_size"] = kwargs.get("batch_size", 512 if "crop" in arch else 2)
+    kwargs["batch_size"] = kwargs.get("batch_size", 512 if model_type == "crop" else 2)
     input_shape = _model.cfg["input_shape"][1:]
     predictor = OrientationPredictor(
         PreProcessor(input_shape, preserve_aspect_ratio=True, symmetric_pad=True, **kwargs),
@@ -60,7 +74,8 @@ def crop_orientation_predictor(
     -------
         OrientationPredictor
     """
-    return _orientation_predictor(arch, load_in_8_bit, engine_cfg, **kwargs)
+    model_type = "crop"
+    return _orientation_predictor(arch, model_type, load_in_8_bit, engine_cfg, **kwargs)
 
 
 def page_orientation_predictor(
@@ -88,4 +103,5 @@ def page_orientation_predictor(
     -------
         OrientationPredictor
     """
-    return _orientation_predictor(arch, load_in_8_bit, engine_cfg, **kwargs)
+    model_type = "page"
+    return _orientation_predictor(arch, model_type, load_in_8_bit, engine_cfg, **kwargs)

{onnxtr-0.4.0 → onnxtr-0.5.0}/onnxtr/models/predictor/base.py

@@ -9,7 +9,7 @@ import numpy as np
 
 from onnxtr.models.builder import DocumentBuilder
 from onnxtr.models.engine import EngineConfig
-from onnxtr.utils.geometry import extract_crops, extract_rcrops, rotate_image
+from onnxtr.utils.geometry import extract_crops, extract_rcrops, remove_image_padding, rotate_image
 
 from .._utils import estimate_orientation, rectify_crops, rectify_loc_preds
 from ..classification import crop_orientation_predictor, page_orientation_predictor
@@ -55,13 +55,19 @@ class _OCRPredictor:
     ) -> None:
         self.assume_straight_pages = assume_straight_pages
         self.straighten_pages = straighten_pages
+        self._page_orientation_disabled = kwargs.pop("disable_page_orientation", False)
+        self._crop_orientation_disabled = kwargs.pop("disable_crop_orientation", False)
         self.crop_orientation_predictor = (
             None
             if assume_straight_pages
-            else crop_orientation_predictor(load_in_8_bit=load_in_8_bit, engine_cfg=clf_engine_cfg)
+            else crop_orientation_predictor(
+                load_in_8_bit=load_in_8_bit, engine_cfg=clf_engine_cfg, disabled=self._crop_orientation_disabled
+            )
         )
         self.page_orientation_predictor = (
-            page_orientation_predictor(load_in_8_bit=load_in_8_bit, engine_cfg=clf_engine_cfg)
+            page_orientation_predictor(
+                load_in_8_bit=load_in_8_bit, engine_cfg=clf_engine_cfg, disabled=self._crop_orientation_disabled
+            )
             if detect_orientation or straighten_pages or not assume_straight_pages
             else None
         )
@@ -112,8 +118,8 @@ class _OCRPredictor:
             ]
         )
         return [
-            # We exapnd if the page is wider than tall and the angle is 90 or -90
-            rotate_image(page, angle, expand=page.shape[1] > page.shape[0] and abs(angle) == 90)
+            # expand if height and width are not equal, afterwards remove padding
+            remove_image_padding(rotate_image(page, angle, expand=page.shape[0] != page.shape[1]))
             for page, angle in zip(pages, origin_pages_orientations)
         ]
 
@@ -123,13 +129,18 @@ class _OCRPredictor:
         loc_preds: List[np.ndarray],
         channels_last: bool,
         assume_straight_pages: bool = False,
+        assume_horizontal: bool = False,
     ) -> List[List[np.ndarray]]:
-        extraction_fn = extract_crops if assume_straight_pages else extract_rcrops
-
-        crops = [
-            extraction_fn(page, _boxes[:, :4], channels_last=channels_last)  # type: ignore[operator]
-            for page, _boxes in zip(pages, loc_preds)
-        ]
+        if assume_straight_pages:
+            crops = [
+                extract_crops(page, _boxes[:, :4], channels_last=channels_last)
+                for page, _boxes in zip(pages, loc_preds)
+            ]
+        else:
+            crops = [
+                extract_rcrops(page, _boxes[:, :4], channels_last=channels_last, assume_horizontal=assume_horizontal)
+                for page, _boxes in zip(pages, loc_preds)
+            ]
         return crops
 
     @staticmethod
@@ -138,8 +149,9 @@ class _OCRPredictor:
         loc_preds: List[np.ndarray],
         channels_last: bool,
         assume_straight_pages: bool = False,
+        assume_horizontal: bool = False,
     ) -> Tuple[List[List[np.ndarray]], List[np.ndarray]]:
-        crops = _OCRPredictor._generate_crops(pages, loc_preds, channels_last, assume_straight_pages)
+        crops = _OCRPredictor._generate_crops(pages, loc_preds, channels_last, assume_straight_pages, assume_horizontal)
 
         # Avoid sending zero-sized crops
         is_kept = [[all(s > 0 for s in crop.shape) for crop in page_crops] for page_crops in crops]

{onnxtr-0.4.0 → onnxtr-0.5.0}/onnxtr/models/predictor/predictor.py

@@ -100,6 +100,8 @@ class OCRPredictor(NestedObject, _OCRPredictor):
             origin_pages_orientations = None
         if self.straighten_pages:
             pages = self._straighten_pages(pages, seg_maps, general_pages_orientations, origin_pages_orientations)
+            # update page shapes after straightening
+            origin_page_shapes = [page.shape[:2] for page in pages]
 
             # forward again to get predictions on straight pages
             loc_preds = self.det_predictor(pages, **kwargs)  # type: ignore[assignment]
@@ -117,6 +119,7 @@ class OCRPredictor(NestedObject, _OCRPredictor):
             loc_preds,  # type: ignore[arg-type]
             channels_last=True,
             assume_straight_pages=self.assume_straight_pages,
+            assume_horizontal=self._page_orientation_disabled,
         )
         # Rectify crop orientation and get crop orientation predictions
         crop_orientations: Any = []

{onnxtr-0.4.0 → onnxtr-0.5.0}/onnxtr/utils/geometry.py

@@ -391,6 +391,26 @@ def rotate_image(
     return rot_img
 
 
+def remove_image_padding(image: np.ndarray) -> np.ndarray:
+    """Remove black border padding from an image
+
+    Args:
+    ----
+        image: numpy tensor to remove padding from
+
+    Returns:
+    -------
+        Image with padding removed
+    """
+    # Find the bounding box of the non-black region
+    rows = np.any(image, axis=1)
+    cols = np.any(image, axis=0)
+    rmin, rmax = np.where(rows)[0][[0, -1]]
+    cmin, cmax = np.where(cols)[0][[0, -1]]
+
+    return image[rmin : rmax + 1, cmin : cmax + 1]
+
+
 def estimate_page_angle(polys: np.ndarray) -> float:
     """Takes a batch of rotated previously ORIENTED polys (N, 4, 2) (rectified by the classifier) and return the
     estimated angle ccw in degrees
@@ -471,7 +491,7 @@ def extract_crops(img: np.ndarray, boxes: np.ndarray, channels_last: bool = True
 
 
 def extract_rcrops(
-    img: np.ndarray, polys: np.ndarray, dtype=np.float32, channels_last: bool = True
+    img: np.ndarray, polys: np.ndarray, dtype=np.float32, channels_last: bool = True, assume_horizontal: bool = False
 ) -> List[np.ndarray]:
     """Created cropped images from list of rotated bounding boxes
 
@@ -481,6 +501,7 @@ def extract_rcrops(
         polys: bounding boxes of shape (N, 4, 2)
         dtype: target data type of bounding boxes
         channels_last: whether the channel dimensions is the last one instead of the last one
+        assume_horizontal: whether the boxes are assumed to be only horizontally oriented
 
     Returns:
     -------
@@ -498,22 +519,88 @@ def extract_rcrops(
         _boxes[:, :, 0] *= width
         _boxes[:, :, 1] *= height
 
-    src_pts = _boxes[:, :3].astype(np.float32)
-    # Preserve size
-    d1 = np.linalg.norm(src_pts[:, 0] - src_pts[:, 1], axis=-1)
-    d2 = np.linalg.norm(src_pts[:, 1] - src_pts[:, 2], axis=-1)
-    # (N, 3, 2)
-    dst_pts = np.zeros((_boxes.shape[0], 3, 2), dtype=dtype)
-    dst_pts[:, 1, 0] = dst_pts[:, 2, 0] = d1 - 1
-    dst_pts[:, 2, 1] = d2 - 1
-    # Use a warp transformation to extract the crop
-    crops = [
-        cv2.warpAffine(
-            img if channels_last else img.transpose(1, 2, 0),
-            # Transformation matrix
-            cv2.getAffineTransform(src_pts[idx], dst_pts[idx]),
-            (int(d1[idx]), int(d2[idx])),
-        )
-        for idx in range(_boxes.shape[0])
-    ]
+    src_img = img if channels_last else img.transpose(1, 2, 0)
+
+    # Handle only horizontal oriented boxes
+    if assume_horizontal:
+        crops = []
+
+        for box in _boxes:
+            # Calculate the centroid of the quadrilateral
+            centroid = np.mean(box, axis=0)
+
+            # Divide the points into left and right
+            left_points = box[box[:, 0] < centroid[0]]
+            right_points = box[box[:, 0] >= centroid[0]]
+
+            # Sort the left points according to the y-axis
+            left_points = left_points[np.argsort(left_points[:, 1])]
+            top_left_pt = left_points[0]
+            bottom_left_pt = left_points[-1]
+            # Sort the right points according to the y-axis
+            right_points = right_points[np.argsort(right_points[:, 1])]
+            top_right_pt = right_points[0]
+            bottom_right_pt = right_points[-1]
+            box_points = np.array(
+                [top_left_pt, bottom_left_pt, top_right_pt, bottom_right_pt],
+                dtype=dtype,
+            )
+
+            # Get the width and height of the rectangle that will contain the warped quadrilateral
+            width_upper = np.linalg.norm(top_right_pt - top_left_pt)
+            width_lower = np.linalg.norm(bottom_right_pt - bottom_left_pt)
+            height_left = np.linalg.norm(bottom_left_pt - top_left_pt)
+            height_right = np.linalg.norm(bottom_right_pt - top_right_pt)
+
+            # Get the maximum width and height
+            rect_width = max(int(width_upper), int(width_lower))
+            rect_height = max(int(height_left), int(height_right))
+
+            dst_pts = np.array(
+                [
+                    [0, 0],  # top-left
+                    # bottom-left
+                    [0, rect_height - 1],
+                    # top-right
+                    [rect_width - 1, 0],
+                    # bottom-right
+                    [rect_width - 1, rect_height - 1],
+                ],
+                dtype=dtype,
+            )
+
+            # Get the perspective transform matrix using the box points
+            affine_mat = cv2.getPerspectiveTransform(box_points, dst_pts)
+
+            # Perform the perspective warp to get the rectified crop
+            crop = cv2.warpPerspective(
+                src_img,
+                affine_mat,
+                (rect_width, rect_height),
+            )
+
+            # Add the crop to the list of crops
+            crops.append(crop)
+
+    # Handle any oriented boxes
+    else:
+        src_pts = _boxes[:, :3].astype(np.float32)
+        # Preserve size
+        d1 = np.linalg.norm(src_pts[:, 0] - src_pts[:, 1], axis=-1)
+        d2 = np.linalg.norm(src_pts[:, 1] - src_pts[:, 2], axis=-1)
+        # (N, 3, 2)
+        dst_pts = np.zeros((_boxes.shape[0], 3, 2), dtype=dtype)
+        dst_pts[:, 1, 0] = dst_pts[:, 2, 0] = d1 - 1
+        dst_pts[:, 2, 1] = d2 - 1
+        # Use a warp transformation to extract the crop
+        crops = [
+            cv2.warpAffine(
+                src_img,
+                # Transformation matrix
+                cv2.getAffineTransform(src_pts[idx], dst_pts[idx]),
+                (int(d1[idx]), int(d2[idx])),
+            )
+            for idx in range(_boxes.shape[0])
+        ]
+
     return crops  # type: ignore[return-value]

{onnxtr-0.4.0 → onnxtr-0.5.0}/onnxtr/utils/vocabs.py

@@ -25,6 +25,7 @@ VOCABS: Dict[str, str] = {
     "hindi_punctuation": "।,?!:्ॐ॰॥॰",
     "bangla_letters": "অআইঈউঊঋএঐওঔকখগঘঙচছজঝঞটঠডঢণতথদধনপফবভমযরলশষসহ়ঽািীুূৃেৈোৌ্ৎংঃঁ",
     "bangla_digits": "০১২৩৪৫৬৭৮৯",
+    "generic_cyrillic_letters": "абвгдежзийклмнопрстуфхцчшщьюяАБВГДЕЖЗИЙКЛМНОПРСТУФХЦЧШЩЬЮЯ",
 }
 
 VOCABS["latin"] = VOCABS["digits"] + VOCABS["ascii_letters"] + VOCABS["punctuation"]
@@ -59,6 +60,9 @@ VOCABS["vietnamese"] = (
 VOCABS["hebrew"] = VOCABS["english"] + "אבגדהוזחטיכלמנסעפצקרשת" + "₪"
 VOCABS["hindi"] = VOCABS["hindi_letters"] + VOCABS["hindi_digits"] + VOCABS["hindi_punctuation"]
 VOCABS["bangla"] = VOCABS["bangla_letters"] + VOCABS["bangla_digits"]
+VOCABS["ukrainian"] = (
+    VOCABS["generic_cyrillic_letters"] + VOCABS["digits"] + VOCABS["punctuation"] + VOCABS["currency"] + "ґіїєҐІЇЄ₴"
+)
 VOCABS["multilingual"] = "".join(
     dict.fromkeys(
         VOCABS["french"]

onnxtr-0.5.0/onnxtr/version.py

@@ -0,0 +1 @@
+__version__ = 'v0.5.0'

{onnxtr-0.4.0 → onnxtr-0.5.0}/onnxtr.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: onnxtr
-Version: 0.4.0
+Version: 0.5.0
 Summary: Onnx Text Recognition (OnnxTR): docTR Onnx-Wrapper for high-performance OCR on documents.
 Author-email: Felix Dittrich <felixdittrich92@gmail.com>
 Maintainer: Felix Dittrich
@@ -225,9 +225,8 @@ Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
 Requires-Python: <4,>=3.9.0
 Description-Content-Type: text/markdown
 License-File: LICENSE
-Requires-Dist: numpy<2.0.0,>=1.16.0
+Requires-Dist: numpy<3.0.0,>=1.16.0
 Requires-Dist: scipy<2.0.0,>=1.4.0
-Requires-Dist: opencv-python<5.0.0,>=4.5.0
 Requires-Dist: pypdfium2<5.0.0,>=4.11.0
 Requires-Dist: pyclipper<2.0.0,>=1.2.0
 Requires-Dist: shapely<3.0.0,>=1.6.0
@@ -240,8 +239,16 @@ Requires-Dist: anyascii>=0.3.2
 Requires-Dist: tqdm>=4.30.0
 Provides-Extra: cpu
 Requires-Dist: onnxruntime>=1.11.0; extra == "cpu"
+Requires-Dist: opencv-python<5.0.0,>=4.5.0; extra == "cpu"
 Provides-Extra: gpu
 Requires-Dist: onnxruntime-gpu>=1.11.0; extra == "gpu"
+Requires-Dist: opencv-python<5.0.0,>=4.5.0; extra == "gpu"
+Provides-Extra: cpu-headless
+Requires-Dist: onnxruntime>=1.11.0; extra == "cpu-headless"
+Requires-Dist: opencv-python-headless<5.0.0,>=4.5.0; extra == "cpu-headless"
+Provides-Extra: gpu-headless
+Requires-Dist: onnxruntime-gpu>=1.11.0; extra == "gpu-headless"
+Requires-Dist: opencv-python-headless<5.0.0,>=4.5.0; extra == "gpu-headless"
 Provides-Extra: html
 Requires-Dist: weasyprint>=55.0; extra == "html"
 Provides-Extra: viz
@@ -257,6 +264,7 @@ Requires-Dist: mypy>=0.812; extra == "quality"
 Requires-Dist: pre-commit>=2.17.0; extra == "quality"
 Provides-Extra: dev
 Requires-Dist: onnxruntime>=1.11.0; extra == "dev"
+Requires-Dist: opencv-python<5.0.0,>=4.5.0; extra == "dev"
 Requires-Dist: weasyprint>=55.0; extra == "dev"
 Requires-Dist: matplotlib>=3.1.0; extra == "dev"
 Requires-Dist: mplcursors>=0.3; extra == "dev"
@@ -276,7 +284,7 @@ Requires-Dist: pre-commit>=2.17.0; extra == "dev"
 [![codecov](https://codecov.io/gh/felixdittrich92/OnnxTR/graph/badge.svg?token=WVFRCQBOLI)](https://codecov.io/gh/felixdittrich92/OnnxTR)
 [![Codacy Badge](https://app.codacy.com/project/badge/Grade/4fff4d764bb14fb8b4f4afeb9587231b)](https://app.codacy.com/gh/felixdittrich92/OnnxTR/dashboard?utm_source=gh&utm_medium=referral&utm_content=&utm_campaign=Badge_grade)
 [![CodeFactor](https://www.codefactor.io/repository/github/felixdittrich92/onnxtr/badge)](https://www.codefactor.io/repository/github/felixdittrich92/onnxtr)
-[![Pypi](https://img.shields.io/badge/pypi-v0.3.2-blue.svg)](https://pypi.org/project/OnnxTR/)
+[![Pypi](https://img.shields.io/badge/pypi-v0.5.0-blue.svg)](https://pypi.org/project/OnnxTR/)
 
 > :warning: Please note that this is a wrapper around the [doctr](https://github.com/mindee/doctr) library to provide a Onnx pipeline for docTR. For feature requests, which are not directly related to the Onnx pipeline, please refer to the base project.
 
@@ -309,8 +317,10 @@ For GPU support please take a look at: [ONNX Runtime](https://onnxruntime.ai/get
 
 ```shell
 pip install "onnxtr[cpu]"
+pip install "onnxtr[cpu-headless]"  # same as cpu but with opencv-headless
 # with gpu support
 pip install "onnxtr[gpu]"
+pip install "onnxtr[gpu-headless]"  # same as gpu but with opencv-headless
 # with HTML support
 pip install "onnxtr[html]"
 # with support for visualization
@@ -356,6 +366,9 @@ model = ocr_predictor(
     # Additional parameters - meta information
     detect_orientation=False,  # set to `True` if the orientation of the pages should be detected (default: False)
     detect_language=False, # set to `True` if the language of the pages should be detected (default: False)
+    # Orientation specific parameters in combination with `assume_straight_pages=False` and/or `straighten_pages=True`
+    disable_crop_orientation=False,  # set to `True` if the crop orientation classification should be disabled (default: False)
+    disable_page_orientation=False,  # set to `True` if the general page orientation classification should be disabled (default: False)
     # DocumentBuilder specific parameters
     resolve_lines=True,  # whether words should be automatically grouped into lines (default: True)
     resolve_blocks=False,  # whether lines should be automatically grouped into blocks (default: False)
@@ -589,8 +602,8 @@ The smallest combination in OnnxTR (docTR) of `db_mobilenet_v3_large` and `crnn_
 |Library                          |FUNSD (199 pages)              |CORD  (900 pages)              |
 |---------------------------------|-------------------------------|-------------------------------|
 |docTR (CPU) - v0.8.1             | ~1.29s / Page                 | ~0.60s / Page                 |
-|**OnnxTR (CPU)** - v0.1.2        | ~0.57s / Page                 | **~0.25s / Page**             |
-|**OnnxTR (CPU) 8-bit** - v0.1.2  | **~0.38s / Page**             | **~0.14s / Page**             |
+|**OnnxTR (CPU)** - v0.4.1        | ~0.57s / Page                 | **~0.25s / Page**             |
+|**OnnxTR (CPU) 8-bit** - v0.4.1  | **~0.38s / Page**             | **~0.14s / Page**             |
 |EasyOCR (CPU) - v1.7.1           | ~1.96s / Page                 | ~1.75s / Page                 |
 |**PyTesseract (CPU)** - v0.3.10  | **~0.50s / Page**             | ~0.52s / Page                 |
 |Surya (line) (CPU) - v0.4.4      | ~48.76s / Page                | ~35.49s / Page                |
@@ -602,7 +615,7 @@ The smallest combination in OnnxTR (docTR) of `db_mobilenet_v3_large` and `crnn_
 |-------------------------------------|-------------------------------|-------------------------------|
 |docTR (GPU) - v0.8.1                 | ~0.07s / Page                 | ~0.05s / Page                 |
 |**docTR (GPU) float16** - v0.8.1     | **~0.06s / Page**             | **~0.03s / Page**             |
-|OnnxTR (GPU) - v0.1.2                | **~0.06s / Page**             | ~0.04s / Page                 |
+|OnnxTR (GPU) - v0.4.1                | **~0.06s / Page**             | ~0.04s / Page                 |
 |EasyOCR (GPU) - v1.7.1               | ~0.31s / Page                 | ~0.19s / Page                 |
 |Surya (GPU) float16 - v0.4.4         | ~3.70s / Page                 | ~2.81s / Page                 |
 |**PaddleOCR (GPU) - no cls - v2.7.3**| ~0.08s / Page                 | **~0.03s / Page**             |
@@ -621,6 +634,16 @@ If you wish to cite please refer to the base project citation, feel free to use
 }
 ```
 
+```bibtex
+@misc{onnxtr2024,
+    title={OnnxTR: Optical Character Recognition made seamless & accessible to anyone, powered by Onnx},
+    author={Felix Dittrich},
+    year={2024},
+    publisher = {GitHub},
+    howpublished = {\url{https://github.com/felixdittrich92/OnnxTR}}
+}
+```
+
 ## License
 
 Distributed under the Apache 2.0 License. See [`LICENSE`](https://github.com/felixdittrich92/OnnxTR?tab=Apache-2.0-1-ov-file#readme) for more information.

{onnxtr-0.4.0 → onnxtr-0.5.0}/onnxtr.egg-info/requires.txt

@@ -1,6 +1,5 @@
-numpy<2.0.0,>=1.16.0
+numpy<3.0.0,>=1.16.0
 scipy<2.0.0,>=1.4.0
-opencv-python<5.0.0,>=4.5.0
 pypdfium2<5.0.0,>=4.11.0
 pyclipper<2.0.0,>=1.2.0
 shapely<3.0.0,>=1.6.0
@@ -14,9 +13,15 @@ tqdm>=4.30.0
 
 [cpu]
 onnxruntime>=1.11.0
+opencv-python<5.0.0,>=4.5.0
+
+[cpu-headless]
+onnxruntime>=1.11.0
+opencv-python-headless<5.0.0,>=4.5.0
 
 [dev]
 onnxruntime>=1.11.0
+opencv-python<5.0.0,>=4.5.0
 weasyprint>=55.0
 matplotlib>=3.1.0
 mplcursors>=0.3
@@ -29,6 +34,11 @@ pre-commit>=2.17.0
 
 [gpu]
 onnxruntime-gpu>=1.11.0
+opencv-python<5.0.0,>=4.5.0
+
+[gpu-headless]
+onnxruntime-gpu>=1.11.0
+opencv-python-headless<5.0.0,>=4.5.0
 
 [html]
 weasyprint>=55.0

{onnxtr-0.4.0 → onnxtr-0.5.0}/pyproject.toml

@@ -31,9 +31,8 @@ dynamic = ["version"]
 dependencies = [
     # For proper typing, mypy needs numpy>=1.20.0 (cf. https://github.com/numpy/numpy/pull/16515)
     # Additional typing support is brought by numpy>=1.22.4, but core build sticks to >=1.16.0
-    "numpy>=1.16.0,<2.0.0",
+    "numpy>=1.16.0,<3.0.0",
     "scipy>=1.4.0,<2.0.0",
-    "opencv-python>=4.5.0,<5.0.0",
     "pypdfium2>=4.11.0,<5.0.0",
     "pyclipper>=1.2.0,<2.0.0",
     "shapely>=1.6.0,<3.0.0",
@@ -49,9 +48,19 @@ dependencies = [
 [project.optional-dependencies]
 cpu = [
     "onnxruntime>=1.11.0",
+    "opencv-python>=4.5.0,<5.0.0",
 ]
 gpu = [
     "onnxruntime-gpu>=1.11.0",
+    "opencv-python>=4.5.0,<5.0.0",
+]
+cpu-headless = [
+    "onnxruntime>=1.11.0",
+    "opencv-python-headless>=4.5.0,<5.0.0",
+]
+gpu-headless = [
+    "onnxruntime-gpu>=1.11.0",
+    "opencv-python-headless>=4.5.0,<5.0.0",
 ]
 html = [
     "weasyprint>=55.0",
@@ -73,6 +82,7 @@ quality = [
 dev = [
     # Runtime
     "onnxruntime>=1.11.0",
+    "opencv-python>=4.5.0,<5.0.0",
     # HTML
     "weasyprint>=55.0",
     # Visualization

{onnxtr-0.4.0 → onnxtr-0.5.0}/setup.py

@@ -9,7 +9,7 @@ from pathlib import Path
 from setuptools import setup
 
 PKG_NAME = "onnxtr"
-VERSION = os.getenv("BUILD_VERSION", "0.4.0a0")
+VERSION = os.getenv("BUILD_VERSION", "0.5.0a0")
 
 
 if __name__ == "__main__":

onnxtr-0.4.0/onnxtr/version.py

@@ -1 +0,0 @@
-__version__ = 'v0.4.0'