API reference

Reading and annotating

Load images and labels into memory.

read_captcha	Read one or more captcha image files into a Captcha object.
Captcha	Container for one or more captcha images.
annotate	Manual / batch annotation of captcha files.

PyTorch dataset plumbing.

CaptchaDataset	Dataset of labeled captcha images.
transform_image	Load and resize image(s) to a fixed height, preserving aspect ratio.
encode_label	Encode a string label into a 1D LongTensor of vocab indices (no blanks).
decode_indices	Decode a CTC index sequence to a string.
pad_collate	Collate function: pads images to the max width and stacks targets.
DEFAULT_VOCAB

CRNN + CTC architecture.

CRNN	Convolutional Recurrent Neural Network for variable-length text recognition.

Fit a model end-to-end.

fit_model	Train a CRNN captcha model on a directory of labeled images.
FitHistory
sequence_accuracy	Fraction of sequences where the full predicted string equals the target.

Predict labels with masking and fixed-length options.

decrypt: predict labels for captcha images using a trained CRNN model.

Save and load local .pt files.

save_model	Persist a CRNN model to disk.
load_model	Load a CRNN model previously saved with :func:`save_model`.

Publish and fetch models from the Hub.

from_pretrained	Download and load a txtcaptcha CRNN from the Hugging Face Hub.
save_pretrained	Write `model.safetensors` + `config.json` (+ optional `README.md`)
push_to_hub	Create `repo_id` if needed and upload weights, config and model card.
DEFAULT_REPO_ID

Fetch labeled captcha datasets published by the project.

download_dataset	Download (and optionally unzip) one of the R captcha datasets.
available_datasets	List of dataset names published in the R captcha release.

Fetch fresh captchas from live sources for evaluation.

download_captchas	Download up to `n` captchas from `source` into `dest_dir`.
available_sources	Sorted list of registered source names.
DownloadError	Raised when a captcha cannot be fetched or parsed from a source.