MartinKlevs
The Meet implementation discussed [here](https://ai.googleblog.com/2020/10/background-features-in-google-meet.html) uses a method called [light wrapping](https://www.imaging-resource.com/news/2016/02/11/create-natural-looking-composite-images-using-light-wrapping-technique) to post-process the output. This removes the halo effect around the segmentation borders.
The output shape in the model card is 265x256x2, however the model that we have here outputs 256x256x1. The model size also differs by a few KB.
Restarting the process could be inconvenient in some scenarios. The simplest way to do this would be to regularly check whether the background file has been replaced.
Calculating multiple slightly shifted segmentation masks would let us make a higher resolution combined mask. The segmentation input would be shifted by a few pixels up or down each time...
The full meet model often leaves small holes in my face. It could be worthwile to add an optional hole filling step.
### Describe the feature you would like to be added. Paraview has the plugin [Surface LIC](https://www.paraview.org/Wiki/ParaView/Line_Integral_Convolution) that enables the use of 2D Line Integral Convolution (LIC) on any surface. LIC...
Hello. I really like this code, and i use it regularly with great results. There have been some recent advances for the BM3D algorithm: [https://doi.org/10.1109/TIP.2020.3014721](https://doi.org/10.1109/TIP.2020.3014721) It would be great to...