captcha-tensorflow

$ python gen_captcha.py -n 1000 -d

$ time python simple_softmax.py
data loaded
train images: 10000. test images: 2000
label_size: 10, image_size: 6000
...
step = 9100, accuracy = 91.10%
step = 9200, accuracy = 91.40%
step = 9300, accuracy = 92.00%
step = 9400, accuracy = 91.40%
step = 9500, accuracy = 91.35%
step = 9600, accuracy = 90.80%
step = 9700, accuracy = 91.60%
step = 9800, accuracy = 91.65%
step = 9900, accuracy = 90.50%
testing accuracy = 91.05%

real2m46.478s
user2m29.704s
sys0m17.828s

$ python softmax_with_log.py

$ tensorboard --logdir=log

$ time python conv_captcha.py
data loaded
train images: 10000. test images: 2000
label_size: 10, image_size: 6000
...
step 9100, training accuracy = 100.00%, testing accuracy = 98.90%
step 9200, training accuracy = 100.00%, testing accuracy = 98.80%
step 9300, training accuracy = 100.00%, testing accuracy = 98.90%
step 9400, training accuracy = 100.00%, testing accuracy = 98.80%
step 9500, training accuracy = 100.00%, testing accuracy = 98.90%
step 9600, training accuracy = 100.00%, testing accuracy = 98.60%
step 9700, training accuracy = 100.00%, testing accuracy = 98.65%
step 9800, training accuracy = 100.00%, testing accuracy = 98.95%
step 9900, training accuracy = 100.00%, testing accuracy = 98.95%
testing accuracy = 99.15%

real	4m44.143s
user	3m40.896s
sys	0m33.492s

{
    "num_per_image": 1,
    "n_train": 1,
    "label_choices": "0123456789",
    "n_test": 1,
    "width": 60,
    "height": 100,
    "label_size": 10
}

$ python gen_captcha.py -h
usage: gen_captcha.py [-h] [-n N] [-t T] [-d] [-l] [-u] [--npi NPI]

optional arguments:
  -h, --help   show this help message and exit
  -n N         number of captchas permutations
  -t T         ratio of test / train.
  -d, --digit  use digits in labels.
  -l, --lower  use lowercase characters in labels.
  -u, --upper  use uppercase characters in labels.
  --npi NPI    number of characters per image.

$ python gen_captcha.py -dl --npi 2 -n 10 -t 0.1
36 choices: 0123456789abcdefghijklmnopqrstuvwxyz
generating 10 groups of captchas in images/char-2-groups-10/train
generating 1 groups of captchas in images/char-2-groups-10/test
write meta info in images/char-2-groups-10/meta.json

$ ls images/char-2-groups-10/train | wc -l
   12600
$ ls images/char-2-groups-10/test/ | wc -l
    1260