Korea Land and Housing Corporation housing floor plans transformer

transform_korea_lh_dataset.py

# Run this script within downloaded .zip from https://www.data.go.kr/data/15037046/fileData.do (August 2019)
# Contains 330 plans in bmp format, Resolution 230-5092.

import os
import cv2
import base64

# Retrieve all files
json_files = []
def get_files(cwd='.'):
	for f in os.listdir(cwd):
		j = cwd + '/' + f # current file
		if os.path.isdir(f):
			get_files(j)
		else:
			if '.json' in j:
				json_files.append(j)

# Convert from json to image
def json_bmp(f, output):
	with open(f, 'r') as o:
		o = o.read()
		b64_name = o.split('base64')[-1].replace(',','').replace('}','')
		if len(b64_name) % 4 != 0:
			b64_name = b64_name[:len(b64_name)-len(b64_name) % 4]
		imgdata = base64.b64decode(b64_name)
		with open(output, 'wb') as ff:
			ff.write(imgdata)

get_files()

# Iterate through all files, and convert to image
out = 'plans/'
if not os.path.isdir(out):
	os.mkdir(out)

k = 1
szmin, szmax = 1e10, 0
for js in json_files:
	w = out + '{0}.bmp'.format(k)
	json_bmp(js, w)
	k+=1
	sz = cv2.imread(w).shape
	szmin = min(szmin, sz[0], sz[1])
	szmax = max(szmax, sz[0], sz[1])

print('Converted {0} files, res {1}-{2}'.format(k-1, szmin, szmax))