def __init__(self, scale=4, num_res_blocks=32, pretrained_weights=None, name=None):
self.scale = scale
self.num_res_blocks = num_res_blocks
self.model = wdsr_b(scale=scale, num_res_blocks=num_res_blocks)
self.model.compile(optimizer=AdamWithWeightsNormalization(lr=0.001), \
loss=self.mae, metrics=[self.psnr])
if pretrained_weights != None:
self.model.load_weights(pretrained_weights)
print("[OK] weights loaded.")
pass
self.data_loader = DataLoader(scale=scale, crop_size=256)
self.pretrained_weights = pretrained_weights
self.default_weights_save_path = '/content/Anime-Super-Resolution/weights/wdsr-b-' + \
str(self.num_res_blocks) + '-x' + str(self.scale) + '.h5'
self.name = name
pass
def __init__(self,
scale=4,
num_res_blocks=32,
pretrained_weights=None,
name=None):
# def __init__(self, scale=4, num_res_blocks=48, pretrained_weights=None, name=None):
self.scale = scale
self.num_res_blocks = num_res_blocks
self.model = wdsr_b(scale=scale, num_res_blocks=num_res_blocks)
self.model.compile(optimizer=AdamWithWeightsNormalization(lr=0.0001),
loss=self.mae,
metrics=[self.psnr])
# self.model.compile(optimizer=AdamWithWeightsNormalization(lr=0.0001), loss=self.ssim_loss, metrics=[self.psnr])
# self.model.compile(optimizer=AdamWithWeightsNormalization(lr=0.001), loss=self.psnr_loss, metrics=[self.psnr])
if pretrained_weights != None:
self.model.load_weights(pretrained_weights)
print("[OK] weights loaded.")
pass
self.data_loader = DataLoader(scale=scale, crop_size=1024)
self.pretrained_weights = pretrained_weights
self.default_weights_save_path = 'experiments/ckpt/wdsr-b-' + \
str(self.num_res_blocks) + '-x' + str(self.scale) + '.h5'
self.name = name
pass
"""
import numpy as np
import tensorflow as tf
import matplotlib.pyplot as plt
from PIL import Image
from keras import backend as K
from keras.losses import mean_absolute_error, mean_squared_error
from keras.models import load_model
from keras.optimizers import Adam
import random
import os
from model import wdsr_a, wdsr_b
from utils import DataLoader
model = wdsr_b(scale=4, num_res_blocks=32)
model.load_weights('./weights/wdsr-b-32-x4.h5')
data_loader = DataLoader(scale=4)
def evaluate_test(model,
setpath='datasets/train',
difficulty='easy',
name='evaluate'):
images = data_loader.search(setpath)
image = random.choice(images)
hr = data_loader.imread(image)
resize = (hr.size[0] // data_loader.scale, hr.size[1] // data_loader.scale)
hidden_scale = random.uniform(1, 3)
radius = random.uniform(1, 3)
def data_generator(path, batch_size=8, input_shape=96, scale=2):
'''data generator for fit_generator'''
fns = os.listdir(path)
n = len(fns)
i = 0
while True:
lrs, hrs = [], []
for b in range(batch_size):
if i == 0:
np.random.shuffle(fns)
fn = fns[i]
fn = os.path.join(path, fn)
lr, hr = utils.pair(fn, input_shape, scale)
lr = utils.normalization(lr)
hr = utils.normalization(hr)
lrs.append(lr)
hrs.append(hr)
i = (i + 1) % n
lrs = np.array(lrs)
hrs = np.array(hrs)
yield lrs, hrs
model = wdsr_b()
model.compile(optimizer='adam',
loss=mean_absolute_error, metrics=[psnr])
model.fit_generator(data_generator('./datasets/train/'),
steps_per_epoch=50,
epochs=1250)