search

YaN9-Y / lafin

LaFIn: Generative Landmark Guided Face Inpainting
145 stars 28 forks source link

i get error when test model #4

Closed redaghanem closed 4 years ago

redaghanem commented 4 years ago

thanks for sharing your code i use google colab and when i type this line to test model

!python test.py --model 2 --checkpoints checkpoints/celeba_models

i get this result

Loading InpaintingModel generator...

start testing...

End Testing

but folder result is empty

i use this config.yml

config.zip

please help me

YaN9-Y commented 4 years ago

Please check the paths of filelist in your config file and try to save the file lists in a path without whitespaces.

redaghanem commented 4 years ago

thanks for reply

i think it's worked after use this

!python test.py --model 3 --checkpoints checkpoints/celeba_models --output checkpoints/results

why 3 make inpainting test i understand from the documentation is 2 for inpainting test

Loading landmark detector... Loading landmark detector complete! Loading InpaintingModel generator...

start testing...

checkpoints/results/landmark_inpaint/joint/195579.png checkpoints/results/landmark_inpaint/joint/196278.png checkpoints/results/landmark_inpaint/joint/196680.png checkpoints/results/landmark_inpaint/joint/200280.png checkpoints/results/landmark_inpaint/joint/200281.png checkpoints/results/landmark_inpaint/joint/200282.png checkpoints/results/landmark_inpaint/joint/200283.png checkpoints/results/landmark_inpaint/joint/200284.png

End Testing

YaN9-Y commented 4 years ago

--model 2: use ground truth landmarks to give inpaint result --model 3: use predicted landmarks to give inpaint result Hope this can help you!

redaghanem commented 4 years ago

Thanks

Can i ask why when i use model 2 with all same files i get this error above the is there is a fill missing

YaN9-Y commented 4 years ago

Model 2 needs to generate ground truth landmark files first, if the landmark files don't exist, the error will be reported. To generate the landmark files, you can see readme for detail. Model 3 can predict landmarks, so it doesn't require landmark files.