search

AliaksandrSiarohin / motion-cosegmentation

Reference code for "Motion-supervised Co-Part Segmentation" paper
Other
656 stars 140 forks source link

Deformed outputs #20

Open amil-rp-work opened 4 years ago

amil-rp-work commented 4 years ago

Hello @AliaksandrSiarohin ! Thanks for sharing such an awesome work.

I have tried the code using the following command python ./motion-cosegmentation/part_swap.py --source_image ./final.jpeg --target_video ./crop.mp4 --checkpoint vox-first-order.pth.tar --config ./motion-cosegmentation/config/vox-256-sem-10segments.yaml --supervised --first_order_motion_model --swap_index 1,2,3,4,5,6,7,8,9,10,11,12,13,14,15
However, I am getting deformed outputs. I have attached the input video - crop.mp4. This is obtained by using crop-video.py from one of your other repos. I have also attached 4 sample images used on the same video and their 4 deformed results. I made sure that crop.mp4 & test*.jpg are all of 256x256 resolution. face_swap_4_examples.zip
Can you please help me with this ?

AliaksandrSiarohin commented 4 years ago

I guess you should crop also the source images.

amil-rp-work commented 4 years ago

Okay so just the face part should be cropped and provided, right? Since it's cropped the resolution won't be 256x256. What are your suggestions, then?

AliaksandrSiarohin commented 4 years ago

Resize to 256x256