split_train_val.py

import os

data_dir = "./data/"
train_dir = "./train/"
val_dir = "./val/"

breed_names = os.listdir(data_dir)
for breed_name in breed_names:
    if not os.path.isdir(train_dir + breed_name):
        os.mkdir(train_dir + breed_name)

    if not os.path.isdir(val_dir + breed_name):
        os.mkdir(val_dir + breed_name)

    curr_breed_path = data_dir + breed_name + "/"
    filenames = os.listdir(curr_breed_path)
    
    for idx, filename in enumerate(filenames):
        if idx < len(filenames) / 5:
            cmd = "cp " + curr_breed_path + filename + " " + val_dir + breed_name + "/"
            os.system(cmd)
        else:
            cmd = "cp " + curr_breed_path + filename + " " + train_dir + breed_name + "/"
            os.system(cmd)