def store_passage(religion, passage, passage_num, line_index, image_url): file_ending = image_url.rpartition('.')[-1] out_filenames = ['%s_%s.%s' % (int(time.time() * 1000), i, file_ending) for i in range(3)] utils.crop_images(image_url, *[os.path.join(config.IMAGE_DIR, f) for f in out_filenames]) record = {'religion':religion, 'passage':passage, 'passage_num':passage_num, 'selected_line':line_index, 'images':out_filenames} db.save(record) print record
def main(): if not args.dir: from download import download_pascal_voc_2012 download_pascal_voc_2012() args.dir = 'data/VOCdevkit/VOC2012/JPEGImages' # load data and crop gt = utils.crop_images(args.dir, args.cropping_size, args.samples) noise = utils.add_poisson_noise_to_images(gt) # scale to [-0.5, 0.5] gt_scaled = gt / 255.0 - 0.5 noise_scaled = noise / 255.0 - 0.5 del gt del noise losses, avg_psnr_list = train(gt_scaled, noise_scaled, args.batch_size, args.learning_rate, args.layers, args.epochs, args.filters, args.save_path) # plot _, ax = plt.subplots(ncols=2) ax[0].plot(losses) ax[1].plot(avg_psnr_list) plt.show()
def store_passage(religion, passage, passage_num, line_index, image_url): # set defaults if no image was found if line_index == '': line_index = -1 if not image_url: image_url = default_image file_ending = image_url.rpartition('.')[-1] out_filenames = ['%s_%s.%s' % (int(time.time() * 1000), i, file_ending) for i in range(3)] # saves three copies of the image to the stored_images directory utils.crop_images(image_url, *[os.path.join(config.IMAGE_DIR, f) for f in out_filenames]) # persists the record to the db record = {'religion':religion, 'passage':passage, 'passage_num':passage_num, 'selected_line':line_index, 'images':out_filenames} try: db.save(record) except Exception, e: print 'An error occurred while saving the record to the db, trying again...' db.save(record)
import utils import configparser if __name__ == '__main__': config = configparser.ConfigParser() config.read("config.py") prefix = config["DEFAULT"]['prefix'] input_dir = prefix + config["PLATE_CROPPER"]['input_dir'] output_dir = prefix + config["PLATE_CROPPER"]['output_dir'] classified_dir = prefix + config["PLATE_CROPPER"]['classified_dir'] utils.crop_images(input_dir, output_dir) utils.utils.split_into_dirs(classified_dir, input_dir, input_dir)
from keras.callbacks import EarlyStopping, ModelCheckpoint from skimage.transform import resize, warp, AffineTransform, rotate from skimage import io, img_as_ubyte from skimage.util import invert from matplotlib import pyplot as plt import random from cv2 import GaussianBlur # get train train data X_train, Y_train = read_train_data() if 1: ix = 2 img = X_train[ix] label = Y_train[ix] X_tf, Y_tf = crop_images(X_train, Y_train) img_tf = X_tf[ix] label_tf = Y_tf[ix] plt.figure(figsize=(8, 8)) plt.subplot(221) plt.title('image') io.imshow(img) plt.subplot(222) plt.title('label') io.imshow(np.squeeze(label)) plt.subplot(223) plt.title('blur') io.imshow(img_tf) plt.subplot(224)