ModNet-App 1.0: Auto Masker
Added 2021-08-07 01:42:33 +0000 UTCHere we go, the app from the previous post.
As seen in the previous examples, this AI try to mask people from photos or videos.
It is based on this git:
https://github.com/ZHKKKe/MODNet
And it have 3 export modes:
RGBA: It generate a PNG with a alpha channel
RGB + Alpha PNG: A regular RGB Png and another PNG with the alpha as RGB
GreenScreen: A Mp4 with the mask as a green color.
It also have two models.
Portrait: Trained on Portraits
Webcam: Trained with people in front of a webcam.
I forgot to change the text on the button from "Interpolate Video" to "Mask Video/Image", but it will most like have some bugs I need to fix, so I fix the text on the next update.
Comments
The upscale area seen to have tons of forks right now, with quite a few models for different inputs that are always evolving, is really hard to keep track. Gonna take a look on this project, may do some simple GUI if the results are pretty good. Thanks
GRisk
2021-08-09 16:15:24 +0000 UTCPretty decent results!
GRisk
2021-08-09 16:13:37 +0000 UTCi have a good idea for another GUI: https://github.com/xinntao/Real-ESRGAN this is a new upscaling tech which seems to work absolutely wonderful. the portable version only contains the NCNN version but not the pytorch version. would be nice to be able to use that on image sequences etc with a gui.
2021-08-09 01:56:39 +0000 UTCThank for this
Buda Dangol
2021-08-08 20:13:14 +0000 UTCNice) My quick test from a .gif file: Source - https://i.imgur.com/KOEWzsY.mp4 Output - https://i.imgur.com/H9TLyv3.mp4
Sergey Tokarev
2021-08-08 07:48:25 +0000 UTCnice job!! this is the Adobe killer right here!
MindChamber
2021-08-07 02:29:29 +0000 UTC
