203 lines
7.3 KiB
Python
203 lines
7.3 KiB
Python
import jittor as jt
|
|
from jittor import init
|
|
from jittor import nn
|
|
import jittor.transform as transform
|
|
import argparse
|
|
import os
|
|
import numpy as np
|
|
import math
|
|
import itertools
|
|
import time
|
|
import datetime
|
|
import sys
|
|
import cv2
|
|
import time
|
|
|
|
from models import *
|
|
from datasets import *
|
|
|
|
import warnings
|
|
warnings.filterwarnings("ignore")
|
|
|
|
jt.flags.use_cuda = 1
|
|
|
|
parser = argparse.ArgumentParser()
|
|
parser.add_argument("--epoch", type=int, default=0, help="epoch to start training from")
|
|
parser.add_argument("--n_epochs", type=int, default=200, help="number of epochs of training")
|
|
parser.add_argument("--dataset_name", type=str, default="facades", help="name of the dataset")
|
|
parser.add_argument("--batch_size", type=int, default=1, help="size of the batches")
|
|
parser.add_argument("--lr", type=float, default=0.0002, help="adam: learning rate")
|
|
parser.add_argument("--b1", type=float, default=0.5, help="adam: decay of first order momentum of gradient")
|
|
parser.add_argument("--b2", type=float, default=0.999, help="adam: decay of first order momentum of gradient")
|
|
parser.add_argument("--decay_epoch", type=int, default=100, help="epoch from which to start lr decay")
|
|
parser.add_argument("--n_cpu", type=int, default=8, help="number of cpu threads to use during batch generation")
|
|
parser.add_argument("--img_height", type=int, default=256, help="size of image height")
|
|
parser.add_argument("--img_width", type=int, default=256, help="size of image width")
|
|
parser.add_argument("--channels", type=int, default=3, help="number of image channels")
|
|
parser.add_argument(
|
|
"--sample_interval", type=int, default=500, help="interval between sampling of images from generators"
|
|
)
|
|
parser.add_argument("--checkpoint_interval", type=int, default=1, help="interval between model checkpoints")
|
|
opt = parser.parse_args()
|
|
print(opt)
|
|
|
|
def save_image(img, path, nrow=10):
|
|
N,C,W,H = img.shape
|
|
if (N%nrow!=0):
|
|
print("save_image error: N%nrow!=0")
|
|
return
|
|
img=img.transpose((1,0,2,3))
|
|
ncol=int(N/nrow)
|
|
img2=img.reshape([img.shape[0],-1,H])
|
|
img=img2[:,:W*ncol,:]
|
|
for i in range(1,int(img2.shape[1]/W/ncol)):
|
|
img=np.concatenate([img,img2[:,W*ncol*i:W*ncol*(i+1),:]],axis=2)
|
|
min_=img.min()
|
|
max_=img.max()
|
|
img=(img-min_)/(max_-min_)*255
|
|
img=img.transpose((1,2,0))
|
|
if C==3:
|
|
img = cv2.cvtColor(img, cv2.COLOR_BGR2RGB)
|
|
cv2.imwrite(path,img)
|
|
|
|
os.makedirs("images/%s" % opt.dataset_name, exist_ok=True)
|
|
os.makedirs("saved_models/%s" % opt.dataset_name, exist_ok=True)
|
|
|
|
# Loss functions
|
|
criterion_GAN = nn.MSELoss()
|
|
criterion_pixelwise = nn.L1Loss()
|
|
|
|
# Loss weight of L1 pixel-wise loss between translated image and real image
|
|
lambda_pixel = 100
|
|
|
|
# Calculate output of image discriminator (PatchGAN)
|
|
patch = (1, opt.img_height // 2 ** 4, opt.img_width // 2 ** 4)
|
|
|
|
# Initialize generator and discriminator
|
|
generator = GeneratorUNet()
|
|
discriminator = Discriminator()
|
|
|
|
if opt.epoch != 0:
|
|
# Load pretrained models
|
|
generator.load("saved_models/%s/generator_last.pkl" % (opt.dataset_name))
|
|
discriminator.load("saved_models/%s/discriminator_last.pkl" % (opt.dataset_name))
|
|
|
|
# Optimizers
|
|
optimizer_G = jt.optim.Adam(generator.parameters(), lr=opt.lr, betas=(opt.b1, opt.b2))
|
|
optimizer_D = jt.optim.Adam(discriminator.parameters(), lr=opt.lr, betas=(opt.b1, opt.b2))
|
|
|
|
# Configure dataloaders
|
|
transforms_ = [
|
|
transform.Resize(size=(opt.img_height, opt.img_width), mode=Image.BICUBIC),
|
|
transform.ImageNormalize(mean=[0.5, 0.5, 0.5], std=[0.5, 0.5, 0.5]),
|
|
]
|
|
dataloader = ImageDataset("../../data/%s" % opt.dataset_name, transforms_=transforms_).set_attrs(
|
|
batch_size=opt.batch_size,
|
|
shuffle=True,
|
|
num_workers=opt.n_cpu,
|
|
)
|
|
val_dataloader = ImageDataset("../../data/%s" % opt.dataset_name, transforms_=transforms_, mode="val").set_attrs(
|
|
batch_size=10,
|
|
shuffle=True,
|
|
num_workers=1,
|
|
)
|
|
|
|
def sample_images(batches_done):
|
|
"""Saves a generated sample from the validation set"""
|
|
real_B, real_A = next(iter(val_dataloader))
|
|
fake_B = generator(real_A)
|
|
img_sample = np.concatenate([real_A.data, fake_B.data, real_B.data], -2)
|
|
save_image(img_sample, "images/%s/%s.png" % (opt.dataset_name, batches_done), nrow=5)
|
|
|
|
warmup_times = -1
|
|
run_times = 3000
|
|
total_time = 0.
|
|
cnt = 0
|
|
|
|
# ----------
|
|
# Training
|
|
# ----------
|
|
|
|
prev_time = time.time()
|
|
|
|
for epoch in range(opt.epoch, opt.n_epochs):
|
|
for i, (real_B, real_A) in enumerate(dataloader):
|
|
# Adversarial ground truths
|
|
valid = jt.ones([real_A.shape[0], 1]).stop_grad()
|
|
fake = jt.zeros([real_A.shape[0], 1]).stop_grad()
|
|
|
|
# ------------------
|
|
# Train Generators
|
|
# ------------------
|
|
# GAN loss
|
|
fake_B = generator(real_A)
|
|
pred_fake = discriminator(fake_B, real_A)
|
|
loss_GAN = criterion_GAN(pred_fake, valid)
|
|
# Pixel-wise loss
|
|
loss_pixel = criterion_pixelwise(fake_B, real_B)
|
|
# Total loss
|
|
loss_G = loss_GAN + lambda_pixel * loss_pixel
|
|
optimizer_G.step(loss_G)
|
|
|
|
# ---------------------
|
|
# Train Discriminator
|
|
# ---------------------
|
|
# Real loss
|
|
pred_real = discriminator(real_B, real_A)
|
|
loss_real = criterion_GAN(pred_real, valid)
|
|
# Fake loss
|
|
pred_fake = discriminator(fake_B.detach(), real_A)
|
|
loss_fake = criterion_GAN(pred_fake, fake)
|
|
# Total loss
|
|
loss_D = 0.5 * (loss_real + loss_fake)
|
|
optimizer_D.step(loss_D)
|
|
|
|
if warmup_times==-1:
|
|
# --------------
|
|
# Log Progress
|
|
# --------------
|
|
|
|
# Determine approximate time left
|
|
batches_done = epoch * len(dataloader) + i
|
|
batches_left = opt.n_epochs * len(dataloader) - batches_done
|
|
time_left = datetime.timedelta(seconds=batches_left * (time.time() - prev_time))
|
|
prev_time = time.time()
|
|
|
|
# Print log
|
|
jt.sync_all()
|
|
if batches_done % 5 == 0:
|
|
sys.stdout.write(
|
|
"\r[Epoch %d/%d] [Batch %d/%d] [D loss: %f] [G loss: %f, pixel: %f, adv: %f] ETA: %s"
|
|
% (
|
|
epoch,
|
|
opt.n_epochs,
|
|
i,
|
|
len(dataloader),
|
|
loss_D.numpy()[0],
|
|
loss_G.numpy()[0],
|
|
loss_pixel.numpy()[0],
|
|
loss_GAN.numpy()[0],
|
|
time_left,
|
|
)
|
|
)
|
|
|
|
# If at sample interval save image
|
|
if batches_done % opt.sample_interval == 0:
|
|
sample_images(batches_done)
|
|
else:
|
|
jt.sync_all()
|
|
cnt += 1
|
|
print(cnt)
|
|
if cnt == warmup_times:
|
|
jt.sync_all(True)
|
|
sta = time.time()
|
|
if cnt > warmup_times + run_times:
|
|
jt.sync_all(True)
|
|
total_time = time.time() - sta
|
|
print(f"run {run_times} iters cost {total_time} seconds, and avg {total_time / run_times} one iter.")
|
|
exit(0)
|
|
if opt.checkpoint_interval != -1 and epoch % opt.checkpoint_interval == 0:
|
|
# Save model checkpoints
|
|
generator.save(os.path.join(f"saved_models/{opt.dataset_name}/generator_last.pkl"))
|
|
discriminator.save(os.path.join(f"saved_models/{opt.dataset_name}/discriminator_last.pkl"))
|