split.py

from glob import glob
import shutil
import random
import os

classes = glob("500x500/*")
outputPath = "output/"
test_percentage = 0.1


n_class = len(classes)
samples2 = []
total = 0
tt = 0
for clss in range(0, len(classes)):
    imgPaths = glob(classes[clss]+"/*")
    tstSam = int(len(imgPaths)*test_percentage)
    
    print("Class {}:\n\tPath: {}\n\tTotal Count: {}\n\tSampled Imgs: {}".format(clss,classes[clss],len(imgPaths),tstSam))
    smp = random.sample(imgPaths, tstSam)
    for pt in smp:
        os.makedirs(outputPath+os.path.basename(classes[clss]), exist_ok=True)
        os.rename(pt, outputPath+os.path.basename(classes[clss])+"/"+os.path.basename(pt))
    samples2.append(smp)
    tt += len(imgPaths)
    total += tstSam
    

print("-"*64 + "\nTotal: {}/{}".format(total,tt))