search

hustvl / Matte-Anything

[Image and Vision Computing (Vol.147 Jul. '24)] Interactive Natural Image Matting with Segment Anything Models
MIT License
465 stars 33 forks source link

Added support for input text and transparency corrections #5

Open neuromorph opened 1 year ago

neuromorph commented 1 year ago

Congratulations and thanks for sharing Matte Anything code.
I made a notebook to try the demo in Google colab and liked it. I then updated the matte_anything code to add support for text input instead of selecting points. The transparency correction settings are included as well. (from the ToDo list).
You can have a look at my fork here. Some samples are added to ReadMe.

JingfengYao commented 1 year ago

Amazing! Thanks a lot for your contribution! 😄 We will check it soon and see whether to put it in our repo. Thanks again for liking Matte Anything.

neuromorph commented 1 year ago

Sure. Happy to contribute. Let me know.