images = [] labels = [] i = 0 sid = 0 # defining classes as dictionary emotions = {'No': 0, 'Yes': 1} for item in image: # getting label corresponding to the image a = data['Answer.Q1Answer'].loc[ data['Input.image_url'] == 'https://lijingyang.me/images/AmazonMTurk/' + item.split('/')[-1]] try: # removing more than one entry image = cv2.imread('/home/stu15/s15/ts6442/Capstone/Labelled_images/' + item.split('/')[-1]) ap = AspectAwarePreprocessor(128, 128) image = ap.preprocess(image) image = img_to_array(image) b = a.values.tolist()[0].split('|')[0] images.append(image) labels.append(emotions[b]) sid += 1 except: # deleting files with label as "NaN" # os.remove('/home/stu15/s15/ts6442/Capstone/Labelled_images/' + item.split('/')[-1]) i += 1 print('[INFO] Exception at image number', sid) pass if sid % 500 == 0: print('[INFO] {} images loaded...'.format(sid)) '''
# construct the argument parser and parse the arguments ap = argparse.ArgumentParser() ap.add_argument('-s', '--size', type=int, default=128, help='desired size of the image') args = vars(ap.parse_args()) # grab the list of images print('[INFO] loading images...') imagePaths = glob.glob('/home/stu15/s15/ts6442/Capstone/images/images/*.jpg') # Resize the image keeping aspect ratio in mind aap = AspectAwarePreprocessor(args['size'], args['size']) # Resize the image without aspect ratio in mind # sp = SimplePreprocessor(128, 128) # converting images to array for easier processing iap = ImageToArrayPreprocessor() # load the dataset from disk then scale the raw pixel intensities to the range [0, 1] sdl = SimpleDatasetLoader(preprocessors=[aap, iap]) # as there are no labels using '_' in place of labels (data, _) = sdl.load(imagePaths, verbose=1000) data = data.astype('float') / 255.0 print('[INFO] total number of images are ', len(imagePaths)) (trainX, testX, _, _) = train_test_split(data, _, test_size=0.1) print('[INFO] train and test split created...') print(trainX.shape)