Exports the object into a nested dict format

Update the state of the metric with new predictions Args: gt: list of groung-truth character sequences pred: list of

Load a set of parameters onto a model >>> from doctr.models import load_pretrained_params >>> load_pretrained_params(model, "https://yoursour

Estimate the angle of the general document orientation based on the lines of the document and the assumption that they should be horizontal.

Builds a convolutional-based layer sequence >>> from torch.nn import Sequential >>> from doctr.models import conv_sequence >>> module = S

Rotate image around the center, interpolation=NEAREST, pad with 0 (black) Args: img: image to rotate geoms: array of geometries o

Crop a set of bounding boxes from an image Args: img_path: path to the image geoms: a array of polygons of shape (N, 4, 2) or of

Read an image file (or a collection of image files) and convert it into an image in numpy format >>> from doctr.io import DocumentFile

Read a PDF file >>> from doctr.io import DocumentFile >>> doc = DocumentFile.from_pdf("path/to/your/doc.pdf") Args:

Overlay the result on a given image Args: interactive: whether the display should be interactive preserve_aspect_rati

Renders the full text of the element

package requirement helper Args: name: name of the package extra_message: additional message to display if the package is no

Load pretrained parameters onto the model Args: path_or_url: the path or URL to the model parameters (checkpoint) **k

Resolves a compatible ImageFont for the system Args: font_family: the font family to use font_size: the size of the font upon ren

Initialize the predictor based on the request Args: request: input request Returns: Callable: the predictor

Computes the aggregated metrics Returns: a dictionary with the exact match score for the raw data, its lower-case counterpart, it

Download a file using its URL >>> from doctr.models import download_from_url >>> download_from_url("https://yoursource.com/yourcheckpoint-you

Invert a list of dict of elements to a dict of list of elements and the other way around Args: x: a list of dictionaries with the same ke

Save model and its configuration on HF hub >>> from doctr.models import login_to_hub, push_to_hf_hub >>> from doctr.models.recognition import

Compute the confidence score for a polygon : mean of the p values on the polygon Args: pred (np.ndarray): p map returned by the m

Crop orientation classification architecture. >>> import numpy as np >>> from doctr.models import crop_orientation_predictor >>> model =

Add positional information to the target sequence and pass it through the decoder.

Text detection architecture. >>> import numpy as np >>> from doctr.models import detection_predictor >>> model = detection_predictor(arch

Export the page as XML (hOCR-format) convention: https://github.com/kba/hocr-spec/blob/master/1.2/spec.md Args: file_titl

Page orientation classification architecture. >>> import numpy as np >>> from doctr.models import page_orientation_predictor >>> model =

Runs command using run_lambda, returns the first regex match if it exists

Generate a synthetic text image Args: text: the text to render as an image font_size: the size of the font font_family: t

Read an image file as a PyTorch tensor Args: npy_img: image encoded as a numpy array of shape (H, W, C) in np.uint8 dtype: the de

Create a matplotlib patch for the element Args: geometry: bounding box (straight or rotated) of the element page_dimensions: dime

Export model to ONNX format. >>> import torch >>> from doctr.models.classification import resnet18 >>> from doctr.models.utils import exp

Execute a given function in parallel for each element of a given sequence >>> from doctr.utils.multithreading import multithread_exec >>> ent

End-to-end OCR architecture using one model for localization, and another for text recognition. >>> import numpy as np >>> from doctr.models

Read an image file as a PyTorch tensor Args: img_path: location of the image file dtype: the desired data type of the output tens

Set the device and dtype of a model and its batches >>> import torch >>> from torch import nn >>> from doctr.models.utils import set_devi

Computes the aggregated metrics Returns: a tuple with the recall, precision and meanIoU scores

Move the predictor to the desired device Args: predictor: the predictor to move Returns: Callable: the predictor moved to th

Sort bounding boxes from top to bottom, left to right Args: boxes: bounding boxes of shape (N, 4) or (N, 4, 2) (in case of rotate

Crop and image and associated bboxes Args: img: image to crop boxes: array of boxes to clip, absolute (int) or relative (float)

Read a byte stream as a PyTorch tensor Args: img_content: bytes of a decoded image dtype: the desired data type of the output ten

Encode character sequences using a given vocab as mapping Args: sequences: the list of character sequences of size N vocab: the o

Convert a list of UploadFile objects to lists of numpy arrays and their corresponding filenames Args: files: list of UploadFile objects

Get languages of a text using langdetect model. Get the language with the highest probability or no language if only a few words or a low probabil

Login to huggingface hub

Expand a polygon (points) by a factor unclip_ratio, and returns a polygon Args: points: The first parameter. Returns:

Text recognition architecture. Example:: >>> import numpy as np >>> from doctr.models import recognition_predictor >>> mo

Fuse batchnorm and conv layers and reparameterize the model Args: model: the FAST model to reparameterize Returns: the repar

Split crops horizontally if they exceed a given aspect ratio. Args: crops: List of image crops (H, W, C). max_ratio: Aspect

Synthesize the page from the predictions Args: **kwargs: keyword arguments passed to the `synthesize_page` method Return

Convert a PIL Image to a PyTorch tensor Args: pil_img: a PIL image dtype: the output tensor data type Returns: decod

Updates the metric Args: gts: a set of relative bounding boxes either of shape (N, 4) or (N, 5) if they are rotated ones

Updates the metric Args: gt_boxes: a set of relative bounding boxes either of shape (N, 4) or (N, 5) if they are rotated ones

Determine how to split the height/width for the cross-shaped windows. Args: size: A tuple (h, w). Returns:

Computes the IoU between two sets of bounding boxes Args: boxes_1: bounding boxes of shape (N, 4) in format (xmin, ymin, xmax, ymax)

Encode a list of gts sequences into a np array and gives the corresponding* sequence lengths. Args: gts: list of ground-t

Convert a geometry to relative coordinates Args: geoms: a set of polygons of shape (N, 4, 2) or of straight boxes of shape (N, 4)

Expands a 2-point line, so that the first is on the edge. In other terms, we extend the line in the same direction until we meet one of the edges.

Export the page as XML (hOCR-format) convention: https://github.com/kba/hocr-spec/blob/master/1.2/spec.md Args: file_titl

Load pretrained parameters onto the model Args: path_or_url: the path or URL to the model parameters (checkpoint) **k

Load pretrained parameters onto the model Args: path_or_url: the path or URL to the model parameters (checkpoint) **k

Get the shape of an image

MobileNetV3-Small architecture as described in `"Searching for MobileNetV3", <https://arxiv.org/pdf/1905.02244.pdf>`_. >>> import torch

MobileNetV3-Small architecture as described in `"Searching for MobileNetV3", <https://arxiv.org/pdf/1905.02244.pdf>`_. >>> import torch

Computes the IoU between two sets of rotated bounding boxes Args: polys_1: rotated bounding boxes of shape (N, 4, 2) polys_2: rot