| import numpy as np |
| from glob import glob |
| import cv2 as cv |
| import sklearn as sk |
| import pandas as pd |
| class utils(): |
| def __init__(self,txtPath,imagePath): |
| self.image,self.txtFeature , self.label = self.loadData(txtPath,imagePath) |
| self.const = None |
|
|
| def split(self): |
| split = sk.modelselection.train_test_split(self.Data,test_size = 0.2) |
| return split |
| def loadData(self,txtPath,imgPath): |
| txtFeatur = pd.DataFrame([[float(t) for t in x.split('\n')[0].split(' ')] for x in open(txtPath).readlines()]) |
| txtFeatur.T.loc[[0,1,2,3],:] |
| label = txtFeatur.loc[:,4] |
| txtFeatur = txtFeatur.T.loc[[0,1,2,3],:].T |
| max = txtFeatur.max() |
| txtFeatur = txtFeatur/max |
| self.const = np.max(label) |
| label = label/self.const |
|
|
|
|
| for i,p in enumerate(glob(imgPath+'/*')): |
| bedroom = [] |
| bathroom = [] |
| frontal = [] |
| kitchen = [] |
| place = p.split('/')[-1].split('.')[0].split('_')[1] |
| if place == 'bedroom': |
| bedroom.append(self.preProcess(cv.imread(p))) |
| elif place == 'bathroom': |
| bathroom.append(self.preProcess(cv.imread(p))) |
| elif place == 'frontal': |
| frontal.append(self.preProcess(cv.imread(p))) |
| else: |
| kitchen.append(self.preProcess(cv.imread(p))) |
| if i%500 == 0: |
| print('[INFO] {}th image loaded'.format(i)) |
| |
| |
| |
| |
| return ([bedroom,bathroom,frontal,kitchen],txtFeatur, label) |
| @staticmethod |
| def preProcess(image): |
| image = cv.resize(image,(128,128)) |
| image = cv.cvtColor(image,cv.COLOR_BGR2RGB) |
| image = image/255.0 |
| return image |
