Image recognition captchas using TensorFlow, no need image segmentation, run on Ubuntu 18.04, Python 3.10, Tensorflow 2.10.0, Numpy 1.23.4
Accuracy is directly related to generated Training data size and the Learning steps number. For example, captcha generator: https://github.com/Gregwar/CaptchaBundle
https://www.continuum.io/downloads#linux
https://github.com/tensorflow/tensorflow
https://pypi.python.org/pypi/captcha/0.1.1
put your own captchas in <current_dir>/data/train_data/ for training, <current_dir>/data/valid_data/ for evaluating and <current_dir>/data/test_data/ for recognize testing, images file name must be label_*.jpg or label_*.png and recommend size 128x48. you can also use default generation:
python captcha_gen_default.py
the result file will be <current_dir>/data/train.tfrecord and <current_dir>/data/valid.tfrecord
python captcha_records.py
train and evaluate neural network on CPU or one single GPU
python captcha_train.py
you can also train over multiple GPUs
python captcha_multi_gpu_train.py
python captcha_eval.py
read captchas from <current_dir>/data/test_data/ for recogition
python captcha_recognize.py
result like this
...
image WFPMX_num552.png recognize ----> 'WFPMX'
image QUDKM_num468.png recognize ----> 'QUDKM'