Hello,
I modified Pix2Pix in order to have a loss style instead of an L1 loss (for theoretical reasons). When I test my loss style between two images of my training set, no problem, it works well but in my network I get the error RuntimeError: Given groups=1, weight of size [64, 3, 3, 3, 3], expected input [1, 2, 256, 256] to have 3 channels, but got 2 channels instead. This is an error that I understand very well but I don’t see where it could come from.
import torch
import torchvision.models as models
from .base_model import BaseModel
from . import networks
from util.util import gram_matrix, get_features, style_loss
class Pix2PixModel(BaseModel):
""" This class implements the pix2pix model, for learning a mapping from input images to output images given paired data.
The model training requires '--dataset_mode aligned' dataset.
By default, it uses a '--netG unet256' U-Net generator,
a '--netD basic' discriminator (PatchGAN),
and a '--gan_mode' vanilla GAN loss (the cross-entropy objective used in the orignal GAN paper).
pix2pix paper: https://arxiv.org/pdf/1611.07004.pdf
"""
@staticmethod
def modify_commandline_options(parser, is_train=True):
"""Add new dataset-specific options, and rewrite default values for existing options.
Parameters:
parser -- original option parser
is_train (bool) -- whether training phase or test phase. You can use this flag to add training-specific or test-specific options.
Returns:
the modified parser.
For pix2pix, we do not use image buffer
The training objective is: GAN Loss + lambda_L1 * ||G(A)-B||_1
By default, we use vanilla GAN loss, UNet with batchnorm, and aligned datasets.
"""
# changing the default values to match the pix2pix paper (https://phillipi.github.io/pix2pix/)
parser.set_defaults(norm='batch', netG='unet_256', dataset_mode='aligned')
if is_train:
parser.set_defaults(pool_size=0, gan_mode='vanilla')
parser.add_argument('--lambda_L1', type=float, default=100.0, help='weight for L1 loss')
return parser
def __init__(self, opt):
"""Initialize the pix2pix class.
Parameters:
opt (Option class)-- stores all the experiment flags; needs to be a subclass of BaseOptions
"""
BaseModel.__init__(self, opt)
# specify the training losses you want to print out. The training/test scripts will call <BaseModel.get_current_losses>
self.loss_names = ['G_GAN', 'G_L1', 'D_real', 'D_fake']
# specify the images you want to save/display. The training/test scripts will call <BaseModel.get_current_visuals>
self.visual_names = ['real_A', 'fake_B', 'real_B']
# specify the models you want to save to the disk. The training/test scripts will call <BaseModel.save_networks> and <BaseModel.load_networks>
self.vgg = models.vgg19(pretrained=True).features
self.vgg.to(self.device)
if self.isTrain:
self.model_names = ['G', 'D']
else: # during test time, only load G
self.model_names = ['G']
# define networks (both generator and discriminator)
self.netG = networks.define_G(opt.input_nc, opt.output_nc, opt.ngf, opt.netG, opt.norm,
not opt.no_dropout, opt.init_type, opt.init_gain, self.gpu_ids)
if self.isTrain: # define a discriminator; conditional GANs need to take both input and output images; Therefore, #channels for D is input_nc + output_nc
self.netD = networks.define_D(opt.input_nc + opt.output_nc, opt.ndf, opt.netD,
opt.n_layers_D, opt.norm, opt.init_type, opt.init_gain, self.gpu_ids)
if self.isTrain:
# define loss functions
self.criterionGAN = networks.GANLoss(opt.gan_mode).to(self.device)
self.criterionL1 = torch.nn.L1Loss()
self.criterionStyleLoss = lambda style_grams, target_features : style_loss(style_grams, target_features)
# initialize optimizers; schedulers will be automatically created by function <BaseModel.setup>.
self.optimizer_G = torch.optim.Adam(self.netG.parameters(), lr=opt.lr, betas=(opt.beta1, 0.999))
self.optimizer_D = torch.optim.Adam(self.netD.parameters(), lr=opt.lr, betas=(opt.beta1, 0.999))
self.optimizers.append(self.optimizer_G)
self.optimizers.append(self.optimizer_D)
def set_input(self, input):
"""Unpack input data from the dataloader and perform necessary pre-processing steps.
Parameters:
input (dict): include the data itself and its metadata information.
The option 'direction' can be used to swap images in domain A and domain B.
"""
AtoB = self.opt.direction == 'AtoB'
self.real_A = input['A' if AtoB else 'B'].to(self.device)
self.real_B = input['B' if AtoB else 'A'].to(self.device)
self.image_paths = input['A_paths' if AtoB else 'B_paths']
def forward(self):
"""Run forward pass; called by both functions <optimize_parameters> and <test>."""
self.fake_B = self.netG(self.real_A) # G(A)
def backward_D(self):
"""Calculate GAN loss for the discriminator"""
# Fake; stop backprop to the generator by detaching fake_B
fake_AB = torch.cat((self.real_A, self.fake_B), 1) # we use conditional GANs; we need to feed both input and output to the discriminator
pred_fake = self.netD(fake_AB.detach())
self.loss_D_fake = self.criterionGAN(pred_fake, False)
# Real
real_AB = torch.cat((self.real_A, self.real_B), 1)
pred_real = self.netD(real_AB)
self.loss_D_real = self.criterionGAN(pred_real, True)
# combine loss and calculate gradients
self.loss_D = (self.loss_D_fake + self.loss_D_real) * 0.5
self.loss_D.backward()
def backward_G(self):
"""Calculate GAN and L1 loss for the generator"""
# First, G(A) should fake the discriminator
fake_AB = torch.cat((self.real_A, self.fake_B), 1)
pred_fake = self.netD(fake_AB)
self.loss_G_GAN = self.criterionGAN(pred_fake, True)
# Second, G(A) = B
style_features = get_features(self.real_B, self.vgg)
style_grams = {layer: gram_matrix(style_features[layer]) for layer in style_features}
target = self.fake_B.requires_grad_(True).to(self.device)
target_features = get_features(self.real_B, self.vgg)
self.loss_G_StyleLoss = self.criterionStyleLoss(style_grams, target_features) * self.opt.lambda_L1
# combine loss and calculate gradients
self.loss_G = self.loss_G_GAN + self.loss_G_StyleLoss
self.loss_G.backward()
def optimize_parameters(self):
self.forward() # compute fake images: G(A)
# update D
self.set_requires_grad(self.netD, True) # enable backprop for D
self.optimizer_D.zero_grad() # set D's gradients to zero
self.backward_D() # calculate gradients for D
self.optimizer_D.step() # update D's weights
# update G
self.set_requires_grad(self.netD, False) # D requires no gradients when optimizing G
self.optimizer_G.zero_grad() # set G's gradients to zero
self.backward_G() # calculate graidents for G
self.optimizer_G.step() # udpate G's weights
And then the functions in util that I use to retrieve the features :
def gram_matrix(tensor):
# get batch_size, depth, height, width of tensor
_, d, h, w = tensor.size()
# reshape so we are multiplying heightand width
tensor = tensor.view(d, h * w)
# calc. gram matrix
gram = torch.mm(tensor, tensor.t())
return gram
def get_features(image, model, layers = None):
""" On passe l'image dans le modèle et obtient des features pour les layers qu'on précise
"""
if layers is None:
layers = {'0' : 'conv1_1',
'5' : 'conv2_1',
'10' : 'conv3_1',
'19': 'conv4_1',
'21': 'conv4_2',
'28': 'conv5_1'}
features = {}
x = image
for name, layer in model._modules.items():
x = layer(x) #passing image through layer
if name in layers:
features[layers[name]] = x
return features
global style_weights
style_weights = {'conv1_1' : 1,
'conv2_1' : 0.75,
'conv3_1': 0.2,
'conv4_1' : 0.2,
'conv5_1' : 0.2}
def style_loss(style_grams, target_features):
loss = 0
for layer in style_weights:
target_feature = target_features[layer]
target_gram = gram_matrix(target_feature)
_, d, h, w = target_feature.shape
style_gram = style_grams[layer]
layer_style_loss = style_weights[layer] * torch.mean((target_gram - style_gram)**2)
loss += layer_style_loss / (d * h * w)
return loss
Thank you in advance for your help