Data Augmentation in TensorFlow Object Detection API

In my pipeline configuration file, i have several data augmentation fields specified, such as:

  data_augmentation_options {
    random_vertical_flip{
    }
  data_augmentation_options {
    random_black_patches {
    }
  }

If I specify the data augmentation options here, are they automatically performed on my training data ? I f so, how? I don’t see it specified anywhere in my code, so was wondering if this is done internally somehow.

Hi @timtam4life,

Yes, if you specify the data augmentation options in your pipeline configuration file, they will be automatically performed on your training data.

Thanks.