YAML Config

YAML file 📃

As you can see, there is many configs that you can modify for your own project, let's have a look.

DATASET:
  DATASET_DIR: './dataset' # Your Dataset Root Path
  TRAINING_DIR: './dataset/train' # Your Training Dataset Path
  TESTING_DIR: './dataset/val' # Your Testing Dataset Path

  TRAINING_DATASET_LEN: 10000 # Your Training Dataset Length
  TESTING__DATASET_LEN: 1000 # Your Testing Dataset Length

  CAPTCHA:
    FONT_DIR: './fonts/font1.ttf'
    IMG_HEIGHT: 60
    IMG_WIDTH: 160
    CLASS_NUM: 36
    CHAR_LEN: 4

SOLVER:
  LR: 5.0e-4
  BATCH_SIZE: 256
  EPOCH: 5

LOGGER:
  CHECKPOINT_DIR: './checkpoint'
  LOG_DIR: './logs'

DATASET

The DATASET field is about basic settings of your DATASET including dataset path and the length of it

CAPTCHA field is more about the CAPTCHA generator, FONT_DIR refers to the font that you want to use to generate the CAPTCHA dataset. IMG_HEIGHT and IMG_WIDTH is the input image size of the model. The CLASS_NUM field refers to the output class number of the model, the default value of it is 36 (10 numbers and 26 lowercases) . The CHAR_LEN is the length of the CAPTCHA.

SOLVER

SOLVER field is related to the network model, now, it has two sub-fields LR learning rate (default=5e-4) and BATCH_SIZE batch size(default=256), you can modify them to do your own experiment.

LOGGER

You can set the logger output dir in LOGGER field, it will tell you where to find the output after you training the model. Remember to check it.

TODO

• Add Model Choice
• Add Epoch Choice