Welcome to crnn.gluon’s documentation!

crnn.gluon

config module

convert_rec module

crnn module

class crnn.BidirectionalGRU(hidden_size, num_layers, nOut)

基类：mxnet.gluon.block.HybridBlock

__init__(hidden_size, num_layers, nOut)

Initialize self. See help(type(self)) for accurate signature.

hybrid_forward(F, x, *args, **kwargs)

Overrides to construct symbolic graph for this Block.

x : Symbol or NDArray

The first input tensor.

*args : list of Symbol or list of NDArray

Additional input tensors.

class crnn.BidirectionalLSTM(hidden_size, num_layers, nOut)

基类：mxnet.gluon.block.HybridBlock

__init__(hidden_size, num_layers, nOut)

Initialize self. See help(type(self)) for accurate signature.

hybrid_forward(F, x, *args, **kwargs)

Overrides to construct symbolic graph for this Block.

x : Symbol or NDArray

The first input tensor.

*args : list of Symbol or list of NDArray

Additional input tensors.

class crnn.CRNN(n_class, hidden_size=256, num_layers=1)

基类：mxnet.gluon.block.HybridBlock

__init__(n_class, hidden_size=256, num_layers=1)

Initialize self. See help(type(self)) for accurate signature.

hybrid_forward(F, x, *args, **kwargs)

Overrides to construct symbolic graph for this Block.

x : Symbol or NDArray

The first input tensor.

*args : list of Symbol or list of NDArray

Additional input tensors.

class crnn.Decoder(n_class, hidden_size=256, num_layers=1)

基类：mxnet.gluon.block.HybridBlock

__init__(n_class, hidden_size=256, num_layers=1)

Initialize self. See help(type(self)) for accurate signature.

hybrid_forward(F, x, *args, **kwargs)

Overrides to construct symbolic graph for this Block.

x : Symbol or NDArray

The first input tensor.

*args : list of Symbol or list of NDArray

Additional input tensors.

class crnn.DenseNet

基类：mxnet.gluon.block.HybridBlock

__init__()

Initialize self. See help(type(self)) for accurate signature.

hybrid_forward(F, x, *args, **kwargs)

Overrides to construct symbolic graph for this Block.

x : Symbol or NDArray

The first input tensor.

*args : list of Symbol or list of NDArray

Additional input tensors.

class crnn.Encoder

基类：mxnet.gluon.block.HybridBlock

__init__()

Initialize self. See help(type(self)) for accurate signature.

hybrid_forward(F, x, *args, **kwargs)

Overrides to construct symbolic graph for this Block.

x : Symbol or NDArray

The first input tensor.

*args : list of Symbol or list of NDArray

Additional input tensors.

class crnn.ResNet

基类：mxnet.gluon.block.HybridBlock

__init__()

Initialize self. See help(type(self)) for accurate signature.

hybrid_forward(F, x, *args, **kwargs)

Overrides to construct symbolic graph for this Block.

x : Symbol or NDArray

The first input tensor.

*args : list of Symbol or list of NDArray

Additional input tensors.

class crnn.VGG

基类：mxnet.gluon.block.HybridBlock

__init__()

Initialize self. See help(type(self)) for accurate signature.

hybrid_forward(F, x, *args, **kwargs)

Overrides to construct symbolic graph for this Block.

x : Symbol or NDArray

The first input tensor.

*args : list of Symbol or list of NDArray

Additional input tensors.

dataset module

class dataset.ImageDataset(data_txt: str, data_shape: tuple, img_channel: int, num_label: int, alphabet: str, phase: str = 'train')

基类：mxnet.gluon.data.dataset.Dataset

__init__(data_txt: str, data_shape: tuple, img_channel: int, num_label: int, alphabet: str, phase: str = 'train')

数据集初始化

参数:

• data_txt – 存储着图片路径和对于label的文件
• data_shape – 图片的大小(h,w)
• img_channel – 图片通道数
• num_label – 最大字符个数,应该和网络最终输出的序列宽度一样
• alphabet – 字母表

label_enocder(label)

对label进行处理，将输入的label字符串转换成在字母表中的索引

参数:label – label字符串 返回:索引列表

pre_processing(img_path)

对图片进行处理，先按照高度进行resize，resize之后如果宽度不足指定宽度，就补黑色像素，否则就强行缩放到指定宽度

参数:img_path – 图片地址 返回:

class dataset.RecordDataset(filename, data_shape: tuple, img_channel: int, num_label: int)

基类：mxnet.gluon.data.dataset.RecordFileDataset

A dataset wrapping over a RecordIO file contraining images Each sample is an image and its corresponding label

__init__(filename, data_shape: tuple, img_channel: int, num_label: int)

Initialize self. See help(type(self)) for accurate signature.

label_enocder(label)

对label进行处理，将输入的label字符串转换成在字母表中的索引

参数:label – label字符串 返回:索引列表

pre_processing(img)

对图片进行处理 :param img_path: 图片 :return:

keys module

predict module

class predict.GluonNet(model_path, alphabet, img_shape, net, img_channel=3, gpu_id=None)

基类：object

__init__(model_path, alphabet, img_shape, net, img_channel=3, gpu_id=None)

初始化gluon模型 :param model_path: 模型地址 :param alphabet: 字母表 :param img_shape: 图像的尺寸(w,h) :param net: 网络计算图，如果在model_path中指定的是参数的保存路径，则需要给出网络的计算图 :param img_channel: 图像的通道数: 1,3 :param gpu_id: 在哪一块gpu上运行

pre_processing(img_path)

对图片进行处理，先按照高度进行resize，resize之后如果宽度不足指定宽度，就补黑色像素，否则就强行缩放到指定宽度 :param img_path: 图片地址 :return:

predict(img_path)

对传入的图像进行预测，支持图像地址和numpy数组 :param img_path: 图像地址 :return:

predict.decode(preds, alphabet, raw=False)

predict.try_gpu(gpu)

If GPU is available, return mx.gpu(0); else return mx.cpu()

train module

train.accuracy(predictions, labels, alphabet)

train.evaluate_accuracy(net, dataloader, ctx, alphabet)

train.setup_logger(log_file_path: str = None)

train.train()

Indices and tables

• 索引
• 模块索引
• 搜索页面