Ecosyste.ms: Issues
An open API service for providing issue and pull request metadata for open source projects.
GitHub / ypwhs/captcha_break issues and pull requests
#73 - 请问我改如何替换掉ctc_2019中的lambda方法呢 因为lambda在加载保存的模型会有错误
Issue -
State: open - Opened by xiaoguaishoubaobao about 2 years ago
- 1 comment
#72 - tensorflow 2.0 训练的时候 日志不显示 不知道 训练到哪一步
Issue -
State: open - Opened by hechengdu over 2 years ago
- 1 comment
#71 - cnn_2019.ipynb(防止 tensorflow 占用所有显存)tensorflow2.0要怎么改
Issue -
State: open - Opened by hechengdu over 2 years ago
- 1 comment
#68 - 训练完了怎么用啊,纯小白
Issue -
State: open - Opened by yiqingkuang almost 3 years ago
- 2 comments
#67 - 效果不理想
Issue -
State: open - Opened by chengpengvb over 3 years ago
#66 - CTC模型不定长输出问题
Issue -
State: open - Opened by ctzhang2008 over 3 years ago
- 1 comment
#65 - 使用CTC, 识别时不限制4个字符长度,识别率如何?
Issue -
State: open - Opened by ctzhang2008 over 3 years ago
- 3 comments
#64 - 请问如果是不定长的验证码 可以使用吗
Issue -
State: open - Opened by AzureSkyHuHu almost 4 years ago
- 2 comments
#63 - 尝试把n_class+1程序可以运行,但是不知道对不对
Issue -
State: open - Opened by shanhedian2017 about 4 years ago
- 1 comment
#62 - train和val的acc都可以到99%,但是eval,只有0.00265,这是怎么回事呢
Issue -
State: open - Opened by shanhedian2017 about 4 years ago
#61 - 运行winpy/main.py遇到的问题
Issue -
State: open - Opened by AzureSkyHuHu about 4 years ago
- 5 comments
#60 - 如果验证码最后两位相同,似乎一定识别错误
Issue -
State: open - Opened by devmonkeyx about 4 years ago
- 4 comments
#59 - 变长标签怎么处理
Issue -
State: closed - Opened by lsh1994 over 4 years ago
#58 - cannot import name '_imaging' from 'PIL'
Issue -
State: open - Opened by xiaoqian9 over 4 years ago
#57 - Update README.md
Pull Request -
State: closed - Opened by lywly over 4 years ago
#56 - loss为负数且不断减小
Issue -
State: open - Opened by jonechoulover over 4 years ago
#55 - 对Pytorch 添加多卡训练支持
Pull Request -
State: open - Opened by cyy53589 over 4 years ago
- 1 comment
#54 - 对Pytorch 添加多卡训练支持。
Pull Request -
State: closed - Opened by cyy53589 over 4 years ago
- 1 comment
#53 - 多行验证码如何识别呢?
Issue -
State: open - Opened by MichaelGou1105 over 4 years ago
- 4 comments
#52 - 楼主您好,请问3500常用汉字的验证码识别,该模型大小够吗?
Issue -
State: open - Opened by niujiahe over 4 years ago
- 5 comments
#51 - 解决Windows下多进程报错
Pull Request -
State: closed - Opened by s0ap00 almost 5 years ago
#50 - RNN分类之后, 在评估处怎么获得每个字符的概率呢
Issue -
State: open - Opened by lijiajun3029 almost 5 years ago
- 1 comment
#49 - 您可否告知下这几个文件是独立运行的吗,如何训练自己的中文验证码呢
Issue -
State: open - Opened by bixueting almost 5 years ago
- 5 comments
Labels: 自定义数据
#48 - 如何把保存下来的ctc模型载入继续训练呢?
Issue -
State: closed - Opened by MgArcher almost 5 years ago
#47 - 预测输出咋有空的
Issue -
State: open - Opened by wuyangdut almost 5 years ago
- 1 comment
#46 - 请问ctc_2019中input_length=16, label_length=4的具体使用方法?
Issue -
State: open - Opened by chajnoven almost 5 years ago
- 8 comments
Labels: base_model 与 model
#45 - 请问如果想使用该模型训练非captcha库内的图片,应该怎么修改呢?
Issue -
State: open - Opened by charfole almost 5 years ago
- 20 comments
Labels: 自定义数据, 数据增强
#44 - python 版本?
Issue -
State: closed - Opened by originme almost 5 years ago
- 1 comment
#43 - 请问关于pytorch版本该如何修改测试自己的数据集?
Issue -
State: open - Opened by chajnoven almost 5 years ago
- 2 comments
Labels: 自定义数据, 调优
#42 - 不能多线程运行?
Issue -
State: open - Opened by wuyangdut almost 5 years ago
- 1 comment
#41 - base_model与model的关系
Issue -
State: closed - Opened by MgArcher almost 5 years ago
- 1 comment
#40 - PNG该如何修改参数?
Issue -
State: closed - Opened by Pinndanyi almost 5 years ago
#39 - 将pytorch_ctc改为py文件时出现问题
Issue -
State: open - Opened by BerryRB almost 5 years ago
- 1 comment
#38 - 楼主,请教
Issue -
State: open - Opened by armstrong1972 almost 5 years ago
- 14 comments
Labels: 调优
#37 - 请问下,谁有ctc训练好的模型共享一下(ctc_pytorch.ipynb)
Issue -
State: closed - Opened by wowowo110 about 5 years ago
- 2 comments
#36 - cnn模型构建的问题
Issue -
State: open - Opened by MgArcher about 5 years ago
- 4 comments
Labels: 模型结构
#35 - https://github.com/aero2a/kape-
Issue -
State: open - Opened by letarg0 about 5 years ago
#34 - 用生成器返回的X_text,和直接读取验证码图片获取的像素数组是不同的么? 得到的结果不同。
Issue -
State: closed - Opened by rhypowang about 5 years ago
#33 - 准确率回调函数
Issue -
State: open - Opened by Dearge about 5 years ago
- 2 comments
#32 - 有没有点选的例子
Issue -
State: open - Opened by obrua about 5 years ago
#31 - 用ctc_2019.ipynb训练的模型,预测时16个通道只有前两个和最后两个会预测,而且准确率很差?
Issue -
State: open - Opened by rhypowang about 5 years ago
- 16 comments
#30 - 不同大小的图片不能一起训练,导致模型泛化能力很弱
Issue -
State: open - Opened by brucefeng10 about 5 years ago
- 9 comments
Labels: 调优
#29 - 在ctc_pytorch的训练过程中,损失值总会变为nan
Issue -
State: open - Opened by AemaH about 5 years ago
- 4 comments
Labels: Loss nan
#28 - ctc_2019.ipynb中,训练完一个epoch后,就不再继续了
Issue -
State: open - Opened by AemaH about 5 years ago
- 5 comments
#27 - 请问:DataLoader worker (pid(s) 2832, 15200, 1128, 7276, 11544, 12964, 9324, 4740, 2284, 5624, 7340, 13672) exited unexpectedly 怎么解决
Issue -
State: open - Opened by airdot about 5 years ago
- 8 comments
#26 - 训练报错
Issue -
State: closed - Opened by songhuaiyu about 5 years ago
- 2 comments
#25 - 您好,关于四层全连接层的问题
Issue -
State: open - Opened by yyqgigray over 5 years ago
- 1 comment
Labels: 模型结构
#24 - 如何用python run
Issue -
State: open - Opened by kingking888 over 5 years ago
- 1 comment
#23 - ctc.ipynb merge([gru_1, gru_1b], mode='sum') 找不到这个函数
Issue -
State: open - Opened by QGB over 5 years ago
- 2 comments
#22 - 如果把模型保存下来,然后重新加载load_model报错
Issue -
State: open - Opened by Times125 over 5 years ago
- 10 comments
#21 - 这个验证码识别平台有人用过吗 http://www.damagou.top 注册就送100提分,我注册试了下,还可以
Issue -
State: closed - Opened by mengpengfei over 5 years ago
#20 - 你好, 关于旋转图片的疑问
Issue -
State: closed - Opened by Haagaau22 over 5 years ago
- 2 comments
#19 - about y_pred[:, 2:, :]
Issue -
State: closed - Opened by RenXiandong888 over 5 years ago
- 1 comment
#18 - input_length 表示 y_pred 的长度,我们这里是15,这个15是怎么来的
Issue -
State: open - Opened by little2Rabbit almost 6 years ago
- 1 comment
#17 - 数据生成函数为什么最后要返回一个np.ones(batch_size)?
Issue -
State: open - Opened by Yogaht almost 6 years ago
- 4 comments
#16 - 你好,cnn模型不加BN层不收敛
Issue -
State: open - Opened by chrisyoung2017 about 6 years ago
- 12 comments
Labels: 调优
#15 - 你好,如果需要训练能识别五个字母的模型 该怎么改动参数?
Issue -
State: open - Opened by ghost about 6 years ago
- 4 comments
#14 - 关于keras某些包的更改
Issue -
State: open - Opened by 1033020837 over 6 years ago
- 2 comments
#13 - 请问,如果验证码长度不是4个,而是变化的,在设计模型的时候应该怎么做?
Issue -
State: closed - Opened by sisyphusking over 6 years ago
- 1 comment
#12 - 小白求问,3点疑惑
Issue -
State: closed - Opened by moses1994 over 6 years ago
- 1 comment
#11 - Model loading Issues
Issue -
State: open - Opened by alvincjin over 6 years ago
- 1 comment
#10 - model variable vs base_model
Issue -
State: open - Opened by roccolocko almost 7 years ago
- 3 comments
Labels: base_model 与 model
#9 - 图片切割
Issue -
State: closed - Opened by bestpredicts almost 7 years ago
- 6 comments
#8 - InvalidArgumentError (see above for traceback): Saw a non-null label (index >= num_classes - 1) following a null label, batch: 75 num_classes: 36 labels: [[Node: ctc/CTCLoss = CTCLoss[ctc_merge_repeated=true, ignore_longer_outputs_than_inputs=false, preprocess_collapse_repeated=false, _device="/job:localhost/replica:0/task:0/cpu:0"](ctc/Log/_97, ctc/ToInt64/_99, ctc/ToInt32_2/_101, ctc/ToInt32_1/_103)]]
Issue -
State: closed - Opened by lxj0276 about 7 years ago
- 3 comments
#7 - Saw a non-null label (index >= num_classes - 1) following a null label
Issue -
State: open - Opened by lxj0276 about 7 years ago
- 5 comments
#6 - Any way to make captcha_break more generic?
Issue -
State: closed - Opened by eromoe over 7 years ago
- 5 comments
#5 - Perform bad with same code.
Issue -
State: closed - Opened by eromoe over 7 years ago
- 1 comment
#4 - 用keras2的api改写后,无法收敛
Issue -
State: open - Opened by camperr over 7 years ago
- 18 comments
Labels: 调优
#3 - Update from the origin
Pull Request -
State: closed - Opened by Randnx over 7 years ago
#2 - 你好,能提供下cnn训练出来的模型文件吗
Issue -
State: closed - Opened by gantoday over 7 years ago
- 2 comments
#1 - Input 0 is incompatible with layer dense_3: expected ndim=2, found ndim=3
Issue -
State: open - Opened by keviswang over 7 years ago
- 3 comments