def __init__(self):
weights_dir = '../weights'
finetune_iterations = 100
seg_remove_mouth = True
seg_batch_size = 2
batch_size = 2
detection_model = os.path.join(weights_dir, 'v2/WIDERFace_DSFD_RES152.pth')
pose_model = os.path.join(weights_dir, 'shared/hopenet_robust_alpha1.pth')
lms_model = os.path.join(weights_dir, 'v2/hr18_wflw_landmarks.pth')
seg_model = os.path.join(weights_dir, 'v2/celeba_unet_256_1_2_segmentation_v2.pth')
reenactment_model = os.path.join(weights_dir, 'v2/nfv_msrunet_256_1_2_reenactment_v2.1.pth')
completion_model = os.path.join(weights_dir, 'v2/ijbc_msrunet_256_1_2_inpainting_v2.pth')
blending_model = os.path.join(weights_dir, 'v2/ijbc_msrunet_256_1_2_blending_v2.pth')
criterion_id_path = os.path.join(weights_dir, 'v2/vggface2_vgg19_256_1_2_id.pth')
criterion_id = VGGLoss(criterion_id_path)
self.model = FaceSwapping(
detection_model=detection_model, pose_model=pose_model, lms_model=lms_model,
seg_model=seg_model, reenactment_model=reenactment_model,
completion_model=completion_model, blending_model=blending_model,
criterion_id=criterion_id,
finetune=True, finetune_save=True, finetune_iterations=finetune_iterations,
seg_remove_mouth=finetune_iterations, batch_size=batch_size,
seg_batch_size=seg_batch_size, encoder_codec='mp4v')
'/data/datasets/ijb-c/ijbc_cropped/ijbc_cropped_r256_cs1.2',
'val_list.txt',
frame_window=1,
ignore_landmarks=True,
same_prob=0.0)
numpy_transforms = [RandomHorizontalFlip(), Pyramids(2)]
tensor_transforms = [ToTensor()]
resolutions = [128, 256]
lr_gen = [1e-4, 4e-5]
lr_dis = [1e-5, 4e-6]
epochs = [24, 50]
iterations = ['20k']
batch_size = [32, 16]
workers = 32
pretrained = False
criterion_id = VGGLoss('../../weights/vggface2_vgg19_256_1_2_id.pth')
criterion_attr = VGGLoss('../../weights/celeba_vgg19_256_2_0_28_attr.pth')
criterion_gan = GANLoss(use_lsgan=True)
generator = MultiScaleResUNet(in_nc=7,
out_nc=3,
flat_layers=(2, 2, 2, 2),
ngf=128)
discriminator = MultiscaleDiscriminator(use_sigmoid=True, num_D=2)
optimizer = partial(optim.Adam, betas=(0.5, 0.999))
scheduler = partial(lr_scheduler.StepLR, step_size=10, gamma=0.5)
reenactment_model = '../results/reenactment/ijbc_msrunet_reenactment_attr_no_seg/G_latest.pth'
seg_model = '../../weights/lfw_figaro_unet_256_segmentation.pth'
lms_model = '../../weights/hr18_wflw_landmarks.pth'
rec_weight = 1.0
gan_weight = 0.1
background_value = -1.0
batch_size = 8
detection_model = os.path.join(weights_dir, 'v2/WIDERFace_DSFD_RES152.pth')
pose_model = os.path.join(weights_dir, 'shared/hopenet_robust_alpha1.pth')
lms_model = os.path.join(weights_dir, 'v2/hr18_wflw_landmarks.pth')
seg_model = os.path.join(weights_dir,
'v2/celeba_unet_256_1_2_segmentation_v2.pth')
reenactment_model = os.path.join(
weights_dir, 'v2/nfv_msrunet_256_1_2_reenactment_v2.1.pth')
completion_model = os.path.join(weights_dir,
'v2/ijbc_msrunet_256_1_2_inpainting_v2.pth')
blending_model = os.path.join(weights_dir,
'v2/ijbc_msrunet_256_1_2_blending_v2.pth')
criterion_id_path = os.path.join(weights_dir,
'v2/vggface2_vgg19_256_1_2_id.pth')
criterion_id = VGGLoss(criterion_id_path)
face_swapping = FaceSwapping(detection_model=detection_model,
pose_model=pose_model,
lms_model=lms_model,
seg_model=seg_model,
reenactment_model=reenactment_model,
completion_model=completion_model,
blending_model=blending_model,
criterion_id=criterion_id,
finetune=True,
finetune_save=True,
finetune_iterations=finetune_iterations,
seg_remove_mouth=finetune_iterations,
batch_size=batch_size,
seg_batch_size=seg_batch_size,