skyduy / cnn_keras Goto Github PK

can you help me?

centers = [((2*i+1)end+(9-2i)*begin)/10 for i in range(5)]
分割语句中的这个，就是寻找5个中间点进行分割然后出来的这个进行计算是什么意思？多谢

from utils import load_single_data

ImportError: cannot import name load_single_data

你好，能说下你分割图片的思路吗？
我手上自己在写的一个程序卡在分割上。
谢谢。

Hello
I have captcha and want train it

but error

result = self.forward(*input, **kwargs)
File "train.py", line 38, in forward
x = x.view(-1, 48 * 6 * 27) # flatten here
RuntimeError: shape '[-1, 7776]' is invalid for input of size 475968

my data has : 45,595 file
CHARS = ['A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'K', 'M',
'N','L','S','I', 'P', 'R', 'T', 'U', 'V', 'W', 'X', 'Y','1','2','3','4','5','6','7','8','9']
And my captcha lenght: 6 character
How I can config file train.py correct
thank

你能用纯c++实现不分割只通过训练识别验证码吗我帮你商业化收入应该还不错

为什么不把训练数据集也打包上传呀

Hi,

how can i calculate that parameters in pytorch branch?

train.py

self.fc1 = nn.Linear(48 * 6 * 27, 360)

what does it mean that parameters (48 * 6 * 27), 360 wheres coming?

ty

https://github.com/aero2a/kape-
This is public captcha for Aero2. Very complicated and many digits.
Is possible generate data using this library https://github.com/letarg0/obrazy

Is possible?

When I run predict_1.py after I have install the utils through pip, but still this problem.

Traceback (most recent call last):
  File "predict_2.py", line 14, in <module>
    from utils import asc_chr
ImportError: cannot import name asc_chr

and I didn't find anything about asc_chr in utils source code.
what's wrong with my solution?

Hi, can you tell me about the dataset ? How is it organized ? Thank you ?

我用keras搭建了一个CNN网络，我在model.fit()中设置了我的validation_data, 用的是vadidation_spilt=0.25
但是我运行的时候发现，training data的accuracy是80%左右，但是一开始我的val_acc就有100%了，而且一直保持不变，我想问一下，可能哪些地方出了问题？

The label.csv needs them

謝謝

Hello,
I' trying to use CNN for sequence to sequence task, I just started, I'm inspired from your code just I have 3 input that I concatenate in the first conv layer but I got error:
layer expected input with ndim=4 but previous layer has output_shape (None, None, 300)
any suggestions???

what it's dependency

Hi,

..../Desktop/captcha/core/train.py
Using TensorFlow backend.
... loading data
total letters: 49990
Traceback (most recent call last):
File "C:/Users/..../Desktop/captcha/core/train.py", line 89, in
weight_folder=r'../models'
File "C:/Users/..../Desktop/captcha/core/train.py", line 52, in train
x_train, y_train, x_test, y_test = prepare_data(pic_folder)
File "C:/Users/..../Desktop/captcha/core/train.py", line 25, in prepare_data
data, label = load_data(folder)
File "C:\Users....\Desktop\captcha\core\utils.py", line 34, in load_data
data[index*5+sub_index, :, :, :] = img / 255
ValueError: could not broadcast input array from shape (40,29,3) into shape (40,40,3)

why i get shape error that's any dataset

感谢分享

还是超星支持给了数据? 6000个训练数据耗了多少时间啊。