Forward pass for training the model. Args: x (torch.Tensor): Input tensor of shape (batch_size, ...). target

Generate shared permutations for the whole batch. This works because the same attention mask can be used for the shorter sequences be

Generate shared permutations for the whole batch. This works because the same attention mask can be used for the shorter sequences be

Generate attention masks given a sequence permutation (includes pos. for bos and eos tokens) :param perm: the permutation sequence. i

Generate attention masks given a sequence permutation (includes pos. for bos and eos tokens) :param perm: the permutation sequence. i

return center point and end point of TCL instance; filter with the char maps;

return center point and end point of TCL instance; filter with the char maps;

return center point and end point of TCL instance; filter with the char maps;

Warp all the function together.

r"""Generate a square mask for the sequence. The masked positions are filled with float('-inf'). Unmasked positions are filled with float(0.0).

return metrics { 'acc': 0, 'norm_edit_dis': 0, }

random crop.

获取训练和验证的数据list :param data_dir: 数据集根目录 :param data_path: 训练的dataset文件列表，每个文件内以如下格式存储 ‘path/to/img\tlabel’ :return:

see RARE paper Fig. 6 (a)

input_feed={self.input_name: image_numpy} :param input_name: :param image_numpy: :return:

input_name = onnx_session.get_inputs()[0].name :param onnx_session: :return:

Remove duplicate and get pos idxs of keep items. The value of keep_blank should be [None, 95].

Remove duplicate and get pos idxs of keep items. The value of keep_blank should be [None, 95].

Get the model settings based on the provided parameters or default values. Args: use_doc_orientation_classify (Union[boo

output_name = onnx_session.get_outputs()[0].name :param onnx_session: :return:

cvtColor.

Initialize ONNX inference model. Args: encoder_path: Path to encoder ONNX model. If None, use default cache directory. decoder_pa

gather_info: [[x, y], [x, y] ...] logits_map: H x W X (n_chars + 1)

jitter.

Launch UniRec ONNX Gradio demo with default configuration. Args: encoder_path: Path to encoder ONNX model (default: auto-download)

Load a CLIP model Parameters ---------- name : str A model name listed by `clip.available_models()`, or the path to a model check

Command-line interface for OpenOCR unified interface

Main function

Main test entry point