CNN_LSTM_CTC_Tensorflow

CNN_LSTM_CTC_Tensorflow copied to clipboard

关于batch normalization层

1

如果想要保存跟新moving_mean和moving_variance的话，好像要写以下代码： extra_update_ops = tf.get_collection(tf.GraphKeys.UPDATE_OPS) with tf.control_dependencies(extra_update_ops): train_op = optimizer.minimize(loss) 但是好像没找到相关内容，想问作者是不是漏了写了？

huyi1989

Some problems aboult this code

12

Infact the "max_stepsize" in this code should't be 64.The "max_stepsize" is equal to 12,which is shrunk from original "image_width"(180) to 180/2/2/2/2 = 12.Remenber the core idea in CRNN+CTC is that...

980044579

Does this OCR work with Arabic text?

1

Hi I wanted to know whether this OCR works with joint Arabic text yet?

Aspiringarabist

这个方案好像很多人用，问下是否有相关的论文可否提示下，最近想研究下这块

1

谢谢

hanghang2333

ValueError: need more than 2 values to unpack

3

use 60*180size to train ,num_classes = 12 then i got it loading train data, please wait--------------------- ('get image: ', 15000) loading validation data, please wait--------------------- ('get image: ', 4999) 2017-11-06...

hset911

1. In validation step, 'lastbatch_err' is missing in sess.run(), which leads to crash. 2. feature shape after cnn should depend on input shape of image

toxic-0518