neuralchen / SimSwap

An arbitrary face-swapping framework on images and videos with one single trained model!
Other
4.41k stars 872 forks source link

Can anyone help? Pretrained network G has fewer layers; The following are not initialized: ['down0', "first_layer',"last_layer','up0‘] #435

Open GoodFanxinran opened 1 year ago

GoodFanxinran commented 1 year ago

I got the strange image when I ran this: python test_one_image.py --name people --Arc_path arcface_model/arcface_checkpoint.tar --pic_a_path crop_224/6.jpg --pic_b_path crop_224/ds.jpg --output_path output/

result

Here is the details:

image

Don`t know the reason. @neuralchen Could you please give me some suggestions if you are avaliable?

KleinXin commented 1 year ago

me too many people have reported this problem

https://github.com/neuralchen/SimSwap/issues/432

https://github.com/neuralchen/SimSwap/issues/247

nghiap commented 1 year ago

Have you got this file ( latestnet_G.pth) in your /checkpoints/people folder? and that it is about 215Mb? It not then your model is not sufficiently trained. You can download the trained model from here: https://github.com/neuralchen/SimSwap/releases, if you do 224, then download the older one (not 512), and extract it to the folder above.

shidalao commented 1 year ago

solved, solution form here