DMFN
DMFN copied to clipboard
How long was the model trained? Can we get more checkpoints?
There is only one checkpoint of the model provided in the github. How long was the model trained up to that checkpoint? Is it the same checkpoint as that from the paper? I tried to run the model with an image of a celebrity and a mask over the eyes, and it does not perform as well as in the paper. Would it be possible to have more checkpoints uploaded? Also, are there two different trained models for face inpainting vs regular image inpainting? Or should the same model work for both?