final_classification_processing.py

# import all required packages
from PIL import Image
import math
from PIL import Image
from scipy import misc
from scipy import ndimage
import pandas as pd
import unittest
import os, sys
import shutil #for copying files
import numpy as np
import shutil
import tqdm as tqdm
from methods_temporal_correction import *
from methods_rules_based_approach import *
#-------------------------------------------------------------------------------

flag_overlapping_TempCorrection = False
batch_size_overlapping = 6
#temp_correction_list=['direct_application','combined_yearly_prediction']
temp_correction_list=['combined_yearly_prediction']
year_list = ['2015','2016','2017','2018','2019','2020']
#year_list = ['2010','2011','2012','2013','2014','2015','2016','2017','2018','2019']
#year_list = ['1999','2000','2001','2002','2003','2004','2005','2006','2007','2008','2009','2010','2011','2012','2013','2014','2015','2016','2017','2018','2019','2020']



print(sys.argv[0])
print()
input_folder = sys.argv[1]     #we have stored monthly predections in a folder named by district name
if input_folder.split('/')[-1]=='':
	input_folder = input_folder[:-1]
district_name = input_folder.split('/')[-1]
if (district_name==""): #if user has put extra "/" at the end of input folder path.
	district_name = input_folder.split('/')[-2]

print('Name of Area/District - ',district_name)



##==========================================Sub folder structure creation ======================================================

print('''
        -------------------------------------------------------
        Creating sub-folders structure and
        Converting .tif files into .png for further processing.  
        -------------------------------------------------------
	''')

total_count=0
for infile in os.listdir(input_folder):
    if infile[-4:] == ".tif":
        total_count = total_count+1
        for year in year_list:
            os.makedirs(input_folder+"/"+year,exist_ok=True)
            if (year in infile):
                shutil.move(input_folder+'/'+infile, input_folder+'/'+year+'/'+infile)
print("Total .tif files found in the root folder - ", total_count)


for year in year_list:
	main_folder = input_folder + '/' + year
	os.makedirs(main_folder+"/pngs",exist_ok=True)
	count = 0
	for infile in os.listdir(main_folder):
		if infile[-4:] == ".tif":                   #reading all tif files in given folder
			im = Image.open(main_folder+"/"+infile)
			im.save(main_folder+"/pngs/"+infile[:-4]+'.png')
			count=count+1
	print(count, ' images found and converted for year ',year )		
print('.tif to .png conversion completed.')

print('''
	-----------------------------------------------
	Segregating yearly median prediction 
	and making them colorful.
	-----------------------------------------------
	''')
'''
Make a year_median folder and copy all the yearly_median_prediction for all years in that folder.

'''
os.makedirs(input_folder+"/results/direct_application",exist_ok=True)
for year in year_list:
	src_dir = input_folder + '/' + year + '/pngs'
	dest_dir = input_folder+"/results/direct_application"
	for infile in os.listdir(src_dir):
		if ('year_median' in infile):
			shutil.copyfile(src_dir+'/'+infile, dest_dir+'/'+district_name+'_prediction_'+year+'.png')

make_images_colorful(input_folder, "direct_application")

##=========================================Rule based approach =======================================================

print('''
	-----------------------------------------------
	Processing for final_year_prediction started.
	 (Please wait. This might take few minutes.)
	-----------------------------------------------
	''')

os.makedirs(input_folder+"/results/combined_yearly_prediction",exist_ok=True)
for year in year_list:
	main_folder = input_folder + '/' + year
	#Find the minimum number of background pixels in the images of all months for this year
	dataset = [ np.asarray(Image.open(main_folder+"/pngs/"+infile)) for infile in os.listdir(main_folder+"/pngs/") ]
	image_dimension = dataset[0].shape
	#print(image_dimension)

	#Initializing the results prediction matrix for a particular year
	results_prediction = np.zeros(image_dimension[0] * image_dimension[1]).reshape(image_dimension)
	#print(results_prediction)

	for i in tqdm.tqdm(range(image_dimension[0]), desc='Progress : '):
	    for j in range(image_dimension[1]):
	        x = [ dataset[k][i][j] for k in range(len(dataset)) ]
	        results_prediction[i,j] = merge_prediction(x)

	print("final_prediction "+year +" - ",np.unique(results_prediction,return_counts=True))
	results_prediction = (Image.fromarray(results_prediction)).convert("L")
	results_prediction.save(input_folder+'/results/combined_yearly_prediction/'+district_name+'_prediction_'+year+'.png')        
print('done!')

print('''
	-----------------------------------------------
	final_year_prediction done.
	Making final_year_prediction colorful for you!
	-----------------------------------------------
	''')   


make_images_colorful(input_folder, "combined_yearly_prediction")

##========================================= Temp Correction procedure =========================================================			

print('''
	---------------------------------------------------
	Combined_yearly_prediction for all years completed.
	Temporal correction begins...
	---------------------------------------------------
	''')  


if flag_overlapping_TempCorrection==True:
	print("Temporal correction is being done with overlapping of two years with batches of "
		+ batch_size_overlapping + ' years')
	Overlapping_TempCorrection_execution(input_folder, district_name, year_list, temp_correction_list, batch_size_overlapping)
else:
	print("Temporal correction is being done for all years together.")
	for folder_name in temp_correction_list:
		TempCorrection(input_folder, district_name, year_list, folder_name)		

print('''
	-------------------------------------------------------
	Making temporal corrected predictions colorful for you!
	-------------------------------------------------------
	''')  
#-----------------------------------------------------
for folder_name in temp_correction_list:
	temp_coorected_folder_name = folder_name + '_temp_corrected'
	make_images_colorful(input_folder, temp_coorected_folder_name)


print('''
	-----------------------------------------------
	 Congratulations! We are done with everything.
	-----------------------------------------------
	''')