distribution.py

import cv2
import os
import glob
import skbuild
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt

img_dir = "data/training_data"  # Enter Directory of all images
data_path = os.path.join(img_dir, '*g')
files = glob.glob(data_path)
data = []
yellow_count =0
magenta_count=0
cyan_count =0
black_count =0
blue_count=0
white_count=0
green_count=0


for f1 in files:
    # print(f1)
    img = cv2.imread(f1)
    img=cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
    cyan_count += np.count_nonzero((img == [0, 255, 255]).all(axis=2))      #urban_land
    yellow_count += np.count_nonzero((img == [255,255, 0]).all(axis=2))     #agriculture
    magenta_count += np.count_nonzero((img == [255, 0,255]).all(axis=2))    #range_land
    green_count += np.count_nonzero((img == [0, 255,0]).all(axis=2))        #forest_land
    blue_count += np.count_nonzero((img == [0, 0, 255]).all(axis=2))        #water
    white_count += np.count_nonzero((img == [255, 255, 255]).all(axis=2))   #barren_land
    black_count += np.count_nonzero((img == [0, 0, 0]).all(axis=2))         #unknown


print("yellow:",yellow_count/1e6)
print("magenta:",magenta_count/1e6)
print("cyan:",cyan_count/1e6)
print("black:",black_count/1e6)
print("white:",white_count/1e6)
print("blue:",blue_count/1e6)
print("green:",green_count/1e6)
n_bins=7

df = pd.DataFrame({'Land Type':['urban_land','agriculture','rangeland','forest_land','water','barren_land','unknown'],
                   'Pixels':[cyan_count,yellow_count,magenta_count,green_count,blue_count,white_count,black_count]})

ax = df.plot.bar(x='Land Type', y='Pixels',  width=1)
plt.show()