backscrub icon indicating copy to clipboard operation
backscrub copied to clipboard

Virtual Video Device for Background Replacement with Deep Semantic Segmentation

Results 45 backscrub issues
Sort by recently updated
recently updated
newest added

TFLite are now up to 2.6.1 (stable) and beyond in testing releases: https://github.com/tensorflow/tensorflow/releases These may offer fixes for newer toolchains (GCC 11+, Clang 12+), in particular on more bleeding edge...

enhancement
good first issue
chore

Having recently discovered that the open source [Jitsi](https://jitsi.org) video conferencing solution offers ML driven background replacement, I thought it would be interesting to round up who else is doing this...

documentation

We have a significant number of command line parameters, path resolver behaviour, background types & sources, and models that can all vary - manually checking is now time consuming &...

enhancement
good first issue
chore

I've recently discovered that as of v3.x OpenCV will silently attempt to use OpenCL (and thus the GPU) if it finds the required libraries on the system (https://opencv.org/opencl/). Unfortunately operating...

documentation
wontfix

I got everything build and installed , but get an error message that the models/segm_full_v679.tflite could not be opened. Any idea? **me@Darkvik:/usr/local/bin$ v4l2-ctl --list-devices** v4l2loopback (platform:v4l2loopback-002): /dev/video2 Integrated Camera: Integrated...

In error messages the case for writing `error` should be consistent. There seem to be places with `ERROR:`, `Error:` and `error:` in the code. Maybe we should decide for one...

enhancement
good first issue
chore

The MLKit selfie segmentation model `models/selfiesegmentation_mlkit-256x256-2021_01_19-v1215.f16.tflite` is pretty good, but it could still be finetuned for office chairs. The model card describes the dataset used to train the model as...

enhancement
help wanted

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.

enhancement
experimental

Never heard of v4l2loopback. Is it hard to make this thing work in windows? Because I want to be able to just go to a computer, any computer, which 99%...

experimental

_Originally posted by @BenBE in https://github.com/floe/backscrub/pull/86#discussion_r645342107_ > I think we can do a somewhat hybrid API: Most functions can be exposed as wrappers with C compatibility and some direct C++...

enhancement
needs use case