Inicio Explorar Axuda
Iniciar sesión
subDesTagesMitExtraKaese
/
BackgroundMattingV2
1
0
Fork 0
Ficheiros Incidencias 0 Pull Requests 0 Wiki

Sen descrición

Árbore: fb5961f70b
Ramas Etiquetas
BackgroundMatti...
ZIP TAR.GZ
subDesTagesMitExtraKaese fb5961f70b moved cv2_frame_to_cuda to inference_webcam %!s(int64=4) %!d(string=hai) anos
dataset d993eaac3e Publish train scripts %!s(int64=4) %!d(string=hai) anos
doc 13cb499ece Improve ONNX compatibility %!s(int64=4) %!d(string=hai) anos
images b14e735b43 Update teaser image %!s(int64=4) %!d(string=hai) anos
model 13cb499ece Improve ONNX compatibility %!s(int64=4) %!d(string=hai) anos
LICENSE d766f5b9c7 Update license %!s(int64=4) %!d(string=hai) anos
README.md d993eaac3e Publish train scripts %!s(int64=4) %!d(string=hai) anos
data_path.py d993eaac3e Publish train scripts %!s(int64=4) %!d(string=hai) anos
export_onnx.py 8860031c3e Update comment %!s(int64=4) %!d(string=hai) anos
export_torchscript.py ee86e2636f Initialize %!s(int64=4) %!d(string=hai) anos
inference_images.py 3d819586bf Combines PR-#26 and #28 (#29) %!s(int64=4) %!d(string=hai) anos
inference_speed_test.py 21702785a8 Add argument option for cpu support %!s(int64=4) %!d(string=hai) anos
inference_utils.py ee86e2636f Initialize %!s(int64=4) %!d(string=hai) anos
inference_video.py 9c9b9592e8 add option to composite onto a target video with --video-target-bgr %!s(int64=4) %!d(string=hai) anos
inference_webcam.py fb5961f70b moved cv2_frame_to_cuda to inference_webcam %!s(int64=4) %!d(string=hai) anos
inference_webcam_ts.py fb5961f70b moved cv2_frame_to_cuda to inference_webcam %!s(int64=4) %!d(string=hai) anos
inference_webcam_ts_compositing.py fb5961f70b moved cv2_frame_to_cuda to inference_webcam %!s(int64=4) %!d(string=hai) anos
requirements.txt ee86e2636f Initialize %!s(int64=4) %!d(string=hai) anos
run.sh d51f89b7e9 add run script %!s(int64=4) %!d(string=hai) anos
runCompositing.sh 77f1bc25c2 added webcam inference with compositing %!s(int64=4) %!d(string=hai) anos
runTS.sh a8c25af64d added webcam inference with torch script %!s(int64=4) %!d(string=hai) anos
train_base.py d993eaac3e Publish train scripts %!s(int64=4) %!d(string=hai) anos
train_refine.py d993eaac3e Publish train scripts %!s(int64=4) %!d(string=hai) anos

README.md

Real-Time High-Resolution Background Matting

Teaser

Official repository for the paper Real-Time High-Resolution Background Matting. Our model requires capturing an additional background image and produces state-of-the-art matting results at 4K 30fps and HD 60fps on an Nvidia RTX 2080 TI GPU.

 

Overview

 

Updates

  • [Mar 06 2021] Training script is published.
  • [Feb 28 2021] Paper is accepted to CVPR 2021.
  • [Jan 09 2021] PhotoMatte85 dataset is now published.
  • [Dec 21 2020] We updated our project to MIT License, which permits commercial use.

 

Download

Model / Weights

Video / Image Examples

Datasets

  • PhotoMatte85
  • VideoMatte240K (We are still dealing with licensing. In the meantime, you can visit storyblocks.com to download raw green screen videos and recreate the dataset yourself.)

 

Demo

Scripts

We provide several scripts in this repo for you to experiment with our model. More detailed instructions are included in the files.

  • inference_images.py: Perform matting on a directory of images.
  • inference_video.py: Perform matting on a video.
  • inference_webcam.py: An interactive matting demo using your webcam.

Notebooks

Additionally, you can try our notebooks in Google Colab for performing matting on images and videos.

Virtual Camera

We provide a demo application that pipes webcam video through our model and outputs to a virtual camera. The script only works on Linux system and can be used in Zoom meetings. For more information, checkout:

Web Demo

Developers in the community has helped us build a web demo. See Community Projects section below.

 

Usage / Documentation

You can run our model using PyTorch, TorchScript, TensorFlow, and ONNX. For detail about using our model, please check out the Usage / Documentation page.

 

Training

Configure data_path.pth to point to your dataset. The original paper uses train_base.pth to train only the base model till convergence then use train_refine.pth to train the entire network end-to-end. More details are specified in the paper.

 

Project members

* Equal contribution.

 

License

This work is licensed under the MIT License. If you use our work in your project, we would love you to include an acknowledgement and fill out our survey.

 

Community Projects

A list of projects built by third-party developers in the community. If you have a project to share, fill out this survey.