def generate_images(forwardNet,
vaeNet,
embeddingNet,
valloader,
name="evaluate"):
dataloader_iterator = iter(valloader)
with torch.no_grad():
img, target = next(dataloader_iterator)
img = img.to(device)
target = target.to(device)
img = img[:args.val_num]
target = target[:args.val_num]
embed_mu, embed_var = embeddingNet(target)
eps = torch.randn_like(embed_mu)
embed_std = torch.exp(0.5 * embed_var)
z = eps * embed_std + embed_mu
predict_img = vaeNet.sample(z)
predict_spectrum = forwardNet(predict_img)
img_result = torch.cat([img, predict_img], dim=0)
save_binary_img(img_result.data,
os.path.join(args.checkpoint, f"{name}_imge.png"),
nrow=args.val_num)
plot_amplitude([target, predict_spectrum], ['input', 'generated'],
save_path=os.path.join(args.checkpoint,
f"{name}_spectrum.png"))
def sample_images(net, name="val"):
with torch.no_grad():
z = torch.randn(args.val_num, args.latent_dim)
z = z.to(device)
sampled = net.sample(z)
save_binary_img(sampled.data,
os.path.join(args.checkpoint, f"{name}.png"),
nrow=args.val_num)
def generate_images(net, valloader, name="val"):
dataloader_iterator = iter(valloader)
with torch.no_grad():
img, spe = next(dataloader_iterator)
img = img.to(device)
recons = net.generate(img)
result = torch.cat([img, recons], dim=0)
save_binary_img(result.data,
os.path.join(args.checkpoint, f"{name}.png"),
nrow=args.val_num)
def sample_images_random(net, name="rand_sample"):
with torch.no_grad():
z = torch.randn(args.val_num, args.latent_dim)
z = z.to(device)
results = net.module.sample(z)
image = results["image"]
specturm = results["specturm"]
save_binary_img(image.data,
os.path.join(args.checkpoint, f"{name}_image.png"),
nrow=args.val_num)
plot_spectrum([specturm], ['spectrum'],
os.path.join(args.checkpoint, f"{name}_spectrum.png"))
def generate_images(net, valloader, name="val"):
dataloader_iterator = iter(valloader)
with torch.no_grad():
img, target = next(dataloader_iterator)
img = img.to(device)
target = target.to(device)
out = net.module.generate(img)
recons = out["recons"]
predict_spectrum = out["predict_spectrum"]
result = torch.cat([img, recons], dim=0)
save_binary_img(result.data,
os.path.join(args.checkpoint, f"{name}_image.png"),
nrow=args.val_num)
plot_amplitude([target, predict_spectrum], ['target', 'predict'],
os.path.join(args.checkpoint, f"{name}_amplitude.png"))
def sample_images_spectrum(net, name="rand_sample"):
dataloader_iterator = iter(valloader)
with torch.no_grad():
img, target = next(dataloader_iterator)
img = img.to(device)
target = target.to(device)
z = torch.randn(args.val_num, args.latent_dim)
z = z.to(device)
results = net.module.sample(z, target)
image = results["image"]
specturm = results["specturm"]
result_img = torch.cat([img, image], dim=0)
save_binary_img(result_img.data,
os.path.join(args.checkpoint, f"{name}_image.png"),
nrow=args.val_num)
plot_spectrum([target, specturm], ['target', 'spectrum'],
os.path.join(args.checkpoint, f"{name}_spectrum.png"))
