Trained models

[luke-777]

Hi,

can you please share the trained models (vgg16_bn_fc_gray_epoch150_bs256.pth,...) ?

Thank you!

[zhaoyuzhi]

Hi,

Could you please refer to https://github.com/zhaoyuzhi/PyTorch-Special-Pre-trained-Models?

Thanks

[luke-777]

Thank you,
which weights do I have to use for following 2 arguments?

parser.add_argument('--global_feature_network_path', type = str, \
       default = './trained_models/vgg16_bn_fc_gray_epoch150_bs256.pth', \
       help = 'the path that contains the pre-trained ResNet model')

parser.add_argument('--perceptual_path', type = str, default = './trained_models/vgg16_pretrained.pth', help = 'the path that contains the pre-trained VGG-16 model')

[zhaoyuzhi]

vgg16_bn_fc_gray_epoch150_bs256.pth is a pre-trained network embedded in SCGAN and you can refer to https://github.com/zhaoyuzhi/PyTorch-Special-Pre-trained-Models

vgg16_pretrained.pth is also a pre-trained network, but it is used to compute perceptual loss. It can be downloaded from torchvision