Open TheScientist1900 opened 4 years ago
Did you happen to get this working? I'm at this same point and dont really want to use the Tensorboard portion, and just want the predictions.
yeah, i have run the CTC file. Howerver, i delete the codes about accuracy. and the name of some variables are changed in the Tensorboard portion part but dont update in the CTC file. You can modify the codes according the error. If you have any problems, you can email me.
------------------ 原始邮件 ------------------ 发件人: "Brandon Dubbs"<notifications@github.com>; 发送时间: 2020年3月5日(星期四) 凌晨4:33 收件人: "Breta01/handwriting-ocr"<handwriting-ocr@noreply.github.com>; 抄送: "545322747"<545322747@qq.com>;"Author"<author@noreply.github.com>; 主题: Re: [Breta01/handwriting-ocr] word_classifier_CTC.ipynb accuracy+padding part (#121)
Did you happen to get this working? I'm at this same point and dont really want to use the Tensorboard portion, and just want the predictions.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.
Whats your contact information? I can't seem to get it to work. Do you have the word_classifier_CTC notebook working on your own dataset or...? I can't seem to get my own data to work.
I use wechat more often. Your own data means the data you create or you download from the website the author provides? I download iam datasets, and if you use the data you create, i am afraid the format will be complex.
---Original--- From: "Brandon Dubbs"<notifications@github.com> Date: Thu, Mar 12, 2020 02:17 AM To: "Breta01/handwriting-ocr"<handwriting-ocr@noreply.github.com>; Cc: "Lucifer-demon"<545322747@qq.com>;"Author"<author@noreply.github.com>; Subject: Re: [Breta01/handwriting-ocr] word_classifier_CTC.ipynb accuracy+padding part (#121)
Whats your contact information? I can't seem to get it to work. Do you have the word_classifier_CTC notebook working on your own dataset or...? I can't seem to get my own data to work.
— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or unsubscribe.
@Breta01 targets = tf.sparse_placeholder(dtype=tf.int32, name='targets')
targets_length = targets.get_shape().as_list()[1] it raises an error the shpae of the tensor in unknown, we can't call as_list function and more details like PAD , test_targets are not initialized.